printf: format %S like GNU (%g, not one decimal)

[MsfPablo]

I was diffing our find against GNU findutils 4.11 across the -printf directives and %S came out differently on every file that isn't a round ratio.

GNU prints the sparseness ratio with %g; we use {:.1}:

$ LC_ALL=C gfind . -printf '%S %s %b %p\n' | sort      $ LC_ALL=C find . -printf '%S %s %b %p\n' | sort
0 224 0 .                                               0.0 224 0 .
1 0 0 ./empty                                           1.0 0 0 ./empty
682.667 6 8 ./sub/b.log                                 682.7 6 8 ./sub/b.log
1365.33 3 8 ./a.txt                                     1365.3 3 8 ./a.txt
2048 2 8 ./sub/deep/c.md                                2048.0 2 8 ./sub/deep/c.md

So %g differs from {:.1} in three ways that all show up here: 6 significant digits rather than 1 decimal place, trailing zeros and a trailing . stripped, and a switch to exponent form outside [1e-4, 1e6).

Rust has no %g, so this adds a small format_g helper and routes %S through it. The zero-length and non-unix branches returned the literal "1.0" and are now "1", matching the ./empty line above.

The unit test's expected values are the output of C's printf("%g", ...) under glibc with LC_ALL=C, including the rounding-changes-the-exponent case (999999.9 -> 1e+06) and the exponent-form case (1234567 -> 1.23457e+06). After the change find -printf '%S %p\n' is byte-identical to GNU on my test tree.

One thing I deliberately left alone: GNU honours LC_NUMERIC for %S (it prints 1365,33 in a comma-decimal locale) while we always emit .. That's a separate matter from the formatting itself, so it felt out of scope here.

[codecov]

Codecov Report

❌ Patch coverage is 94.28571% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 92.05%. Comparing base (1f19cdd) to head (f4f13ff).
⚠️ Report is 13 commits behind head on main.

Files with missing lines	Patch %	Lines
src/find/matchers/printf.rs	94.28%	1 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #842      +/-   ##
==========================================
+ Coverage   91.93%   92.05%   +0.12%     
==========================================
  Files          35       35              
  Lines        7251     7353     +102     
  Branches      378      381       +3     
==========================================
+ Hits         6666     6769     +103     
+ Misses        443      441       -2     
- Partials      142      143       +1     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[codspeed-hq]

Merging this PR will not alter performance

✅ 20 untouched benchmarks

---

_{Comparing MsfPablo:fix-printf-sparseness (f4f13ff) with main (9e778f1)}

[github-actions]

Commit f4f13ff has test result changes:

bfs testsuite:

Test results comparison:
  Current:   TOTAL: 312 / PASSED: 266 / FAILED: 40 / SKIPPED: 6
  Reference: TOTAL: 314 / PASSED: 266 / FAILED: 42 / SKIPPED: 6

Changes from main branch:
  TOTAL: -2
  PASSED: +0
  FAILED: -2

Test improvements (2):
  + gnu/files0_from_ok
  + gnu/okdir_path_empty

[sylvestre]

why only unix ?