wallet of satoshi has lots of channel capacity so I guess it's good for increasing overall throughput and usability. It also has fully self-custodial option.

I know that fees on acinq are outrageous, sometimes it'd be cheaper to send on-chain btc (esp at 1sat/vb rates) than via LN, which puts into question the whole model.

Personally the best lightning wallet I used is shockwave (not affiliated or anything, I just tried all options including breeze and zeus):

This is silly and proves nothing. No reasonable person who is against removing op_return limits claims that "spam" can be prevented completely. The use of filters mitigate, not eliminate spam.

No Peter, this isn't valid reason. Critea is ZK rollup, aka they work on using bitcoin as data layer. Even just one glance at their main slogan "making it possible to build everything on Bitcoin" tells you just as much. The people who oppose relaxing op code limits are against this very idea and want bitcoin to be used purely as monetary system. They are justified with this example -- this is EXACTLY what they don't want.

Personally, I don't really have a strict opinion on this matter if I am honest, although I do observe what's happening to Ethereum, which not long ago advertised itself as "world's decentralised computer" (again, aka blockchain used for data). What bothers me though, how the bitcoin core devs are dealing with this drama (which stirs more drama). Even Antoine in the original article calls them "parasitic" ffs.

Bitcoin core is implementation of bitcoin protocol, it's the most widely used and maintained one (only real alternative being bitcoin knots). Bitcoin core is directly descendent from code written by Satoshi, it serves as the standard for bitcoin nodes and wallets (this is key, core dominates node usage -- remember for later), maintained by a global team of contributors.

There are bunch of differences between core and knots, the critical being core is more conservative (typically) and reaches broad consensus (again, typically) while knots offers more advanced tools, like mempool filters and mining options. Both nodes follow bitcoin consensus rules and can be run by anyone including any apps that use bitcoin software.

Recently, a lot of discussion has been around removing limits on OP codes (particularly OP_RETURN). This was brought up by a lot of users in GitHub repo of bitcoin core as well as other social media.

In principle, what removing those limits means is making it easier to "put jpegs on chain". Up until this moment, anyone wanting jpegs (or any data for that matter) on bitcoin, had to use ways around those limits. But now, the bitcoin core developers are removing them. The argument is that people who want data on bitcoin find ways anyway while (example: ordinals, runes). They are the ones who are willing to pay extra fees (like those the jpeg people). Thus this change helps maintain fee market on bitcoin and they are subject to consensus with fees dictating which txs are processed anyway.

What it boils down to (this is a bit subjective but I offer my take here) is whether you subscribe to bitcoin being purely money layer (with limits on what is being put on the blockchain) or data layer (as long as it fits into block size limits according to consensus). It's kinda philosophical what you want bitcoin to be with trade-offs of helping maintain miner fees.

Importantly, this is is all done by bitcoin core devs only, so limited number of people. Back in the day, when there was a "war" on block size, it was waged on another level because the potential change was done through fork (so all nodes running bitcoin had to vote/agree to change or not). Here, the bitcoin core devs are going as far as booting users commenting from GitHub repo. Take from that what you will. Consensus on bitcoin doesn't change but the impact by limited number of people who maintain this piece of software has factually quite big implications.