create file retries: do not sleep before giving up on the last try

[ThomasWaldmann]

In _process_any's retry handler (from #7351), the sleep happens before checking whether this was the last try. When all retries are used up, the error is re-raised anyway, so that final sleep is pointless — and it is the longest one of the backoff sequence: retry 9 sleeps 10^4.5 ms ≈ 31.6 s. On a source with many permanently failing files (e.g. a disk returning I/O errors), that adds ~31.6 s per failing file on top of the earlier backoff sleeps.

Check last_try first and re-raise immediately; only sleep (and log the retry warning) when another attempt will actually follow. The if retry < MAX_RETRIES - 1 / else split around the warning becomes unnecessary, since the handler now only reaches that point when a retry follows.

No behavior change for retried files: same backoff sleeps, same warning messages.

🤖 Generated with Claude Code

[codecov]

Codecov Report

❌ Patch coverage is 33.33333% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 85.14%. Comparing base (88f470e) to head (9bd2690).
⚠️ Report is 4 commits behind head on master.
✅ All tests successful. No failed tests found.

Files with missing lines	Patch %	Lines
src/borg/archiver/create_cmd.py	33.33%	1 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #9874      +/-   ##
==========================================
+ Coverage   85.12%   85.14%   +0.01%     
==========================================
  Files          93       93              
  Lines       15459    15479      +20     
  Branches     2337     2343       +6     
==========================================
+ Hits        13159    13179      +20     
  Misses       1599     1599              
  Partials      701      701              

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