bisect: --find-good: fix the docs

[badp]

Looking at test_bisect_run_with_args, fake-bisector is being called with --require-file=c (= exit code 0 iff the file c is present), but with --find-good "status 0 means bad and other non-zero statuses mean good" (or so claims jj bisect run --help).

So, when evaluating commit c adding file c, fake-bisector finds the file c, has an exit code 0, and that should make the commit bad, not good.

However, the overall behavior of jj is self-consistent: jj calls the commits with the file 'good," the rest of the commits "bad," then declares that the comment that added the file was the first "good" commit. And that's what you'd expect to see from the incantation: jj bisect run --find-good fake-bisector --require-file=c should behave like it does in the test.

The bug, then, is in the documentation. It's not true that --find-good inverts how COMMAND is interpreted: COMMAND is still expected to have exit codes 0 for good commits, and non-zero for bad commits. It's the expectations about the edges of the range that are flipped.

This commit fixes the docs accordingly.

Checklist

If applicable:

• I have updated CHANGELOG.md
• I have updated the documentation (README.md, docs/, demos/)
• I have updated the config schema (cli/src/config-schema.json)
• I have added/updated tests to cover my changes
• I fully understand the code that I am submitting (what it does,
  how it works, how it's organized), including any code drafted by an LLM.
• For any prose generated by an LLM, I have proof-read and copy-edited with
  an eye towards deleting anything that is irrelevant, clarifying anything
  that is confusing, and adding details that are relevant. This includes,
  for example, commit descriptions, PR descriptions, and code comments.

[josephlou5]

LGTM, but still needs maintainer review.