s3: treat UploadPart success without ETag as retryable error - #9823

What does this change do?

Fixes a panic in the S3 multipart upload path reported in #9822: when a
backend (or a proxy / load balancer in front of it, e.g. MinIO behind
sidekick) answers an UploadPart request with HTTP 200 but no ETag
header, s3ChunkWriter.WriteChunk dereferences uout.ETag unconditionally
in a debug log line and crashes the whole ~85 GB transfer with a nil pointer
panic — even though nothing load-bearing needed the ETag at that point.

Instead of only guarding the log line, this treats an ETag-less successful
response as what it actually is — an unusable part — and returns a retryable
error from inside the pacer callback:

if uout == nil || uout.ETag == nil {
    return true, fmt.Errorf("UploadPart response for chunk %d has no ETag", chunkNumber+1)
}
return false, nil

The pacer then retries that chunk with the usual backoff, which both removes
the panic and gives transient proxy glitches a chance to heal themselves.
This follows the direction suggested in the issue report.

Note on behaviour without this change beyond the crash itself: with only a
log-line guard, the nil ETag would be recorded via addCompletedPart and the
transfer would proceed all the way to CompleteMultipartUpload, which would
then fail server-side (InvalidPart) — so retrying at the chunk level is the
correct recovery point.

Linked issue

Fixes #9822

(The issue contains the reporter's own root-cause analysis against v1.75.0,
which I verified is still accurate on master: addCompletedPart does not
deref, the following fs.Debugf line does.)

For new or changed backends

No new backend; small bug fix in the existing s3 backend. Not run:
go run ./fstest/test_all -backends :s3 needs live S3 credentials which I do
not have. Local verification instead:

• go vet ./backend/s3/ — clean
• go test ./backend/s3/ -count=1 — ok
• go build ./... — clean

Checklist

• This change is trivial OR it has been discussed and agreed in the linked issue.
• I have read the contribution guidelines.
• (If I used AI tools to help write this code) I have read and understood the AI-assisted contributions guidance, and I have tested and take ownership of this change myself.
• I have added tests for all changes in this PR if appropriate.
• I have added documentation for the changes if appropriate.
• All commit messages are in house style.
• (Backend changes only) test_all passes for this backend and if submitting a new backend can provide a test account for the integration tester - see CONTRIBUTING.md.
• This Pull Request is ready for review.

AI-assistance disclosure: this PR was prepared with AI assistance (code
location and fix drafted by an agent, verified locally by me as described
above). I have read rclone's AI-assisted contributions guidance and take full
ownership of this change.

Notes: checklist items for tests/docs/test_all are intentionally unticked —
no new tests added (the failure mode requires an S3-compatible endpoint that
omits ETag on 200 responses; happy to add one if maintainers point me at
the right harness seam), docs unaffected, and test_all needs live S3
credentials.