el-remph

> * This should be one commit, not three. Ah my apologies, that's my bad habit of committing WIP as I go so I remember what order I've done things...

> - Doing terminal capability detection in two passes seems needlessly inefficient. Could we merge the sixel test with get_pixel_per_cell? I don't think so, get_pixel_per_cell uses XTWINOPS `CSI Ps ;...

In the absence of any further comments on the clutter of the commit history, I'll just push any further commits and potentially clean up history later (I know `clean up...

Alright, I've squashed the commits from up til now, and the second commit adds the separate variables for `enable_inline_image_config`, and `INLINE_IMAGE_AUTODETECT` to the option table in rc.c. The original `enable_inline_image`...

Sorry I was away, here's a mockup that seems to work, but I'm not sure if this does optimise anything after all, because to know if we need pixel/char we...

>When I wrote "We can live without that for now", I meant it. Wrote the commit in response to @rkta [here](https://github.com/tats/w3m/pull/302#issuecomment-2380636680), but yes, I agree; rolling back as planned

Rebased with detection of ghostty, which has had support for kitty images since very early on afaict. Anyone know the status of this pr?

Rebased with the changes recommended: * Primary devattr querying moved to terms.c, reading from tty into a dynamically resized buffer * Test for sixel before testing for mlterm OSC5379, to...

> - Obscure, but SyncTERM returns \033[= to primary DA instead of \033[?; might as well support it @bptato Yep that's obscure alright (: Done that and also skipped processing...

>In general, small features/changes like this should be squashed into a single commit ah yes, sorry about that, this was before I had discovered the joys of the force push...