Yes introducing new op_codes is risky, hence the years and years and years of development on them.

The 2 op_codes created that make LN possible for example were in development for years before going live, but without them we wouldn't have LN. CTV for example has been in review for 4+ years with a 5btc+ bug bounty on it that's never been claimed and without CTV or something like it we'll never have feature rich and efficient covenants without which would make scaling custody much more difficult.

Lightning has been steadily growing by thousands of percent over the past 5 years and will continue to grow and get better and easier to use, however it's still very early days though and like you I wouldn't actually expect a regular person to understand channels in the slightest so there's still more work to do there.

Luckily there are a bunch of scaling solutions that bitcoin will use because like we've known for many centuries already a global money requires many more layers than 2 to function equally as well for everyone.

Your points are 100% valid. I'll say this, a lot of people who use LN today do so without understanding the technical nuances. But to those who do, the challenges it presents can be smoothened today. We just need a ton of people who need to understand, to a degree, the depth of knowledge needed to appreciate the urgency of this upgrade. You don't want a problem to present itself, before debating a fix, as things move slow in bitcoin soft fork proposals. This enhances the protocol on a level that, it still makes it even more superior than other alternatives, and choosing to refuse any fork of LNhance, is a sensible opinion, but one that needs re-examining once one fully groks the concept of the need for the fork.