file-search: emit "Binary file <path> matches" for binary content

[BackSlasher]

Follows up on #9740 (merged).

Behavior

Mirrors git grep.

A file is considered binary if any of the first 8KB is \0.

• Default: match on a binary file emits Binary file <path> matches (no per-line output). --name-only still prints just the path.
• -a / --text: bypass the check; treat all files as text and emit raw bytes for matches.
• -I / --no-binary: skip binary files entirely, even if they contain matches.

-a and -I are mutually exclusive (clap-enforced).

The heuristic is a local inline check for now (content.iter().take(8192).any(|b| *b == 0)).
Will follow up with unifying with jj_lib::eol::is_binary and lib/src/diff_presentation/mod.rs:71.

Conflicts: if any add-side of a conflict looks binary, the whole file is treated as binary and follows the same three modes above. Per-side awareness is left as a follow-up alongside the existing conflict-side TODO in the code.

Testing

• Match on binary file emits Binary file <path> matches.
• --name-only on binary match prints just the path.
• No output when the pattern doesn't match, even on binary.
• -a bypasses the check and emits raw bytes.
• -I skips binary files entirely.
• -a combined with -I errors out (clap-enforced).
• Binary conflict (any add-side has a null byte) emits Binary file <path> matches.
• --name-only on binary conflict prints just the path.
• -I on binary conflict skips it entirely.

Not in scope

• Streaming file reads: the current code read_alls the whole file into memory before the 8KB binary probe runs, so -I doesn't save any I/O. A streaming reader could probe the first 8KB and short-circuit without reading the rest. Deferred to keep this PR small.
• Unifying is_binary across the codebase (eol, diff_presentation, and now this file).
• .gitattributes binary/-text support. Depends on a wider stack of .gitattributes PRs (design: cornerstone for gitattributes features #7164, Gitattributes filter design #8144, Gitattributes EOL conversion support #8266, gitattributes filter support #8719, git-lfs: add ability to ignore files based on git attributes #9635) that haven't landed.

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.

[PhilipMetzger]

I generally don't think we should follow Git's patterns in our implementation of the equivalent features, they're more like a suggestion of what we shouldn't do or should learn from.

[BackSlasher]

Thanks for looking
Figured copying from git is a good start as mentioned in #8318. If there's a strong preference for other behavior, happy to adapt