feat: Add mtime as optional ETag algorithm in file backend alongside default md5

[ciscorn]

What kind of change does this PR introduce?

Fix #612

What is the current behavior?

#612

In the file backend, ETags are generated by calculating the MD5 hash of the entire file contents, which can take several seconds or even more than 10 seconds for larger files. This becomes particularly problematic when utilizing Range requests.

What is the new behavior?

The ETag in the file backend is now generated using only a file's modification time (mtime) and size. I believe this is sufficient as it follows the same approach used by Nginx (and moreover, the file backend is primarily used in local development).

[fenos]

Hi @ciscorn Awesome optimisation!
Could you please keep this backwards compatible and use an environment variable to control the checksum algorithm please

[ciscorn]

@fenos Thanks!
I’ve modified the implementation to respect the environment variable. While I think “mtime” is a reasonable default, I’ve set “md5” as the default to maintain backward compatibility. What do you think?

[fenos]

Awesome job @ciscorn
Just one little change requested, to keep the md5 as the default algorithm to not break existing projects.

In a major version we could consider switching to this as default

[fenos]

Awesome job @ciscorn 🎉

[github-actions]

🎉 This PR is included in version 1.17.0 🎉

The release is available on GitHub release

Your semantic-release bot 📦🚀