fix: fallback behaviour added to atomic writes

[suleman1412]

Description

Implemented a fallback behaivour for when the parent directory is not writeable, then falls back to $TMPDIR

Added tests to check :

1. NamedTempFile behaivour when created in the same as the parent directory
2. To check the prefix of the generated tempfile
3. To check the file is created in $TMPDIR when the parent is not writeable, implemented permissions control and tested to check the file is deleted propoerly
   Fixes atomic writes in pixi assume the pixi.toml path is writeable #5724

How Has This Been Tested?

AI Disclosure

• [ x ] This PR contains AI-generated content.
  • [ x ] I have tested any AI-generated content in my PR.
  • [ x ] I take responsibility for any AI-generated content in my PR.

Tools: Claude

Checklist:

• [ x ] I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• [ x ] I have added sufficient tests to cover my changes.
• I have verified that changes that would impact the JSON schema have been made in schema/model.py.

[baszalmstra]

You have to check whether this actually works. Not just of the temp file is in the right directory. The assumption is that if the directory is not writable you still can overwrite the file. We need to check this.

[suleman1412]

@baszalmstra persist fails in a readonly directory. In such cases we can fallback to using fs:write() instead, which wouldnt be non atomic but should work.

[kratsg]

@baszalmstra persist fails in a readonly directory. In such cases we can fallback to using fs:write() instead, which wouldnt be non atomic but should work.

Note the previous behavior was fine (I'd imagine the fallback should be to the original behavior that existed in pixi before the PR that introduced atomic writes). Perhaps with a warning to indicate lack of atomicity?

[baszalmstra]

Yep, that sounds good to me!

[suleman1412]

hi @baszalmstra, if directory is readonly then we fallback to use write along with a warning about the loss of atomicity. Addded e2e tests for the same as well, please have a look.

[baszalmstra]

Cant we just return an error from this function and catch the "permission denied error".