Suppress errors on EBADF when unlocking files

[sfc-gh-kstewart]

Description

This error is harmless and happens regularly when delete_after_read is set. This is because we acquire the lock right at the start of the ReadToEnd function and then defer the unlock, but that function also performs the delete. So, by the time it returns and the defer runs the file descriptor is no longer valid.

Testing

Tested manually on a laptop running MacOS. When using acquire_fs_locks in conjunction with delete_after_read I observe this error in the logs before applying this change:

2024-10-08T14:30:23.626-0700	error	reader/reader_unix.go:28	Failed to unlock	{"kind": "receiver", "name": "filelog", "data_type": "logs", "component": "fileconsumer", "path": "<redacted>", "error": "bad file descriptor"}
github.com/open-telemetry/opentelemetry-collector-contrib/pkg/stanza/fileconsumer/internal/reader.(*Reader).unlockFile
	github.com/open-telemetry/opentelemetry-collector-contrib/pkg/stanza@v0.111.0/fileconsumer/internal/reader/reader_unix.go:28
github.com/open-telemetry/opentelemetry-collector-contrib/pkg/stanza/fileconsumer/internal/reader.(*Reader).ReadToEnd
	github.com/open-telemetry/opentelemetry-collector-contrib/pkg/stanza@v0.111.0/fileconsumer/internal/reader/reader.go:126
github.com/open-telemetry/opentelemetry-collector-contrib/pkg/stanza/fileconsumer.(*Manager).consume.func1
	github.com/open-telemetry/opentelemetry-collector-contrib/pkg/stanza@v0.111.0/fileconsumer/file.go:160

Performing the same test with this changed applied eliminates the error.