Agent instructions and test improvements

[dfaure-kdab]

• Improve grid_with_model_in_row testcase (visually)
• Improve run_tests.sh to only run the appropriate test executable(s)
• Instruct LLM agents on how to generate rust code from a slint file.
  Just like humans, they don't find this obvious e.g. the .rs file
  generated for a unittest only has the test harness, and a slint!
  macro, which doesn't allow investigating the generated code.

[expenses]

Some of these comments seem valuable for human commenters too. Could they go in the development/building guide?

[dfaure-kdab]

Some of these comments seem valuable for human commenters too. Could they go in the development/building guide?

Aren't humans (hacking on slint itself) able to read docs/agents/*.md too?

I guess we could link some of those files from docs/development.md ?

[expenses]

Aren't humans (hacking on slint itself) able to read docs/agents/*.md too?

I'm not sure we want to make actual people feel 2nd class compared to machines.

[dfaure-kdab]

Aren't humans (hacking on slint itself) able to read docs/agents/*.md too?

I'm not sure we want to make actual people feel 2nd class compared to machines.

LOL I didn't realize there were politics and egos involved ;-)

What if we simply rename agents/ to contributors/ or developers/? And have something like the last section of AGENTS.md in development.md, to point to all this? (without the "Load when working on" which is really the only agent-specific thing I can see)

[tronical]

Sounds good to me to move these files into directories with more human exposure:). That doesn't have to be in those PR though:)

[dfaure-kdab]

Sounds good to me to move these files into directories with more human exposure:). That doesn't have to be in those PR though:)

Yep, I'll make another PR for that.