I will say at the outset, in the interest of full disclosure, that I am not one of the people who thinks ground weapons in Aurora are frustrating or out of balance for the most part. Nevertheless I think this is an interesting experiment worth looking at.
I know that it's the change most players might perceive as too much, yet I think 1/10 of tonnage share (that is now nearly optimal) is too, too low for a supply really, especially when you can produce a supply before even designing combat units, and keep it with no storage cost. So, if I'll have to deliver more supply units comparing to combat ones to maintain major assault effectively - I'll be just content and satisfied. I have to consider now if I want my different formations to try to break / thin out enemy forces within first days of battle - or I want them to hold out making less shots, yet being more consistent and capable of waiting for reinforcements.
Here I have agreement about the principle but disagreement about the approach. It is certainly true that supply requirements are too low compared to IRL formations, however there is one catch with raising these requirements (or more accurately the tonnage of those requirements) which is that making such a large part of the formation of LOG elements tends to get a lot of them shot at while reducing the combat effectiveness of the formation. This is not unrealistic, rather a bit of a limitation in Aurora that we are limited in how well we can simulate the echelon placement of different elements while maintaining useful formation sizes (given that 20,000 tons is really around the necessary formation size for large planetary invasion combat, any veteran player will know this, but this is also regiment or brigade size and thus a lot of granularity is abstracted into a monolithic formation).
Since 1.12 we now have the option to split out a large part of the LOG elements into rear echelon formations which can be set as replenishment formations, and only put ~5 days of LOG in the forward formations. This has two implications for ground unit design. The major implication is that GSP usage can probably be increased by a factor of 5x to 8x, such that 10% to 15% of a front line formation should be integrated LOG elements and this is enough to remain in supply for a five-day construction cycle (when reinforcements are triggered). The minor implication is that LVH+LOG are now almost completely useless aside from flavor; this could be fixed by doubling the GSP capacity of the large LOG element and marking it as LVH only, which I would recommend including in your own changes for testing. In previous discussions this turns out to give both infantry and vehicle logistics their own pros and cons which is a good balance space to be in.
I think this also implies that either GSP requirements need to be raised across the board, or LOG capacity decreased, the effect is the same either way as long as the component tonnage remains the same so I won't argue one way or the other.
Turning to this work I think there is some imbalance mainly in that the AV weapons are now too cheap in terms of supply. While this might be necessary for balance I do not think it is "believable" personally, large AV guns should be firing fairly expensive shots even if the rate of fire is not high, mainly in comparison to the CAP components. A typical early game MBT is VEH+CAP/MAV and I do not think the major supply consumption should be from the CAP component. I am open to being shown wrong by well-sourced figures but intuitively this seems incorrect to me. However the greater expense of artillery I think is a good change.
I would suggest to start a reform of the supply system by staying close to the formula used by Steve, which works fairly well IMO aside from the overall GSP demand being too low, and handling the increase in artillery supply consumption by increasing the number of shots as you have done but staying within the formula - or perhaps multiplying by a small factor of perhaps 2x at most if this seems too low, since you also want to lower damage (though I have a comment about that later). This would push MB from 18 GSP to 72 which is less than you have here but is still markedly more than (current) MAV or HCAP by a large margin and would still make artillery relatively much more supply hungry, but the magnitude of change is not too extreme. Then, tune actual supply tonnage by global adjustments (e.g. reduce the GSP per LOG ton). I would maintain this point for all of the changes made here, so I will not repeat it alongside other comments but I do think that generally sticking to Steve's approach for the most part is the preferable option.
# Slight tinkering of personal weapons - more like aligning flavour and meaninfullness
This one I personally dislike how it has been done as it imposes a bit of restriction on flavor where I prefer open interpretation by the player. Mainly, the shift of PWI to be "marksman" weapons I think it limiting - what if I prefer to use PWI to indicate the SAW or grenade launcher in an infantry rifle squad? The way PWI works now better supports a range of RP potentials, including marksman weapons if desired (and I've also used this convention in some settings).
I also prefer the standard PW to be one shot with 1/1 base stats, this I think is useful flavor wise and for players to conceive of the ground combat mechanics as it sets a baseline which is easy to understand. The change here, other than doubling the GSP (which I think is entirely arbitrary and unnecessary), at least has no real change on the unit performance and I understand why it was done, but I think in this case having an easily understood 1/1/1 baseline is for gameplay benefit worth the perhaps slight quibble in terms of flavor.
# An increase of Anti-Armour Weapons penetration - it's quite important for me to make AI forces non-helpless against armour-only avalanches
It is reasonable, though with the boost in effectiveness I think the GSP of AV weapons is too cheap compared to others. Invincible armor on UHVs is too much for the game, at least against NPRs, but armor should still be useful and making AV weapons more expensive can accomplish this balance-wise.
# More or less flavouring alteration of Autocannons
Probably my favorite change here. I always treat AC as siege guns or bunker busters as their stats really fit this role well, so the change makes them a bit more effective in this role while making HAC more viable.
I have decreased Damage parameter for such weapons in the same time, just to contain Collateral Damage per Effective Damage, that is currently making artillery nearly useless strategically (I have largerly sacrificed arty's ability to knock out armoured vehicles yet)
This was earlier in the OP but I address it last as my comment is fairly minor: I think it is worth not thinking about collateral damage right now until we have 1.14 (2.0?) in our hands, as there is an 80% reduction of collateral damage as one of the changes. This could make artillery worth using strategically but we will need to actually playtest to know for sure. However the lower damage and higher shots are reasonable so I don't think any change is invalidated - other than the a bit extraneous supply consumption as discussed above.
I doubt I can find time to do any playtesting unfortunately but I will be interested to see what results if any come from this. I have plans eventually to do a heavily modded campaign and would not be opposed to adding a ground combat rework into those changes if the results are encouraging.