fix(datetime.dst): pass Gregorian to tzinfo.dst()

[5j9]

Standard timezone libraries and databases expect standard Gregorian datetime objects to resolve DST offsets correctly. Passing a Jalali datetime object (self) causes two major issues:

1. It results in silent bugs because the database evaluates rules for the Jalali year (e.g., 1405 AD instead of the converted 2026 AD).
2. It can raise type/attribute errors if strict type-checking is enforced.

This converts the instance to its Gregorian representation using self.togregorian() before passing it to tzinfo.dst().

Also:

• Removes the outdated TODO comment on the dst method.
• Adds unit tests using zoneinfo to verify DST calculations for both aware and naive jdatetime instances.

[slashmili]

@5j9 Thanks! Any chance we could backport this to v6?

[slashmili]

I meant to version v5!

[5j9]

@5j9 Thanks! Any chance we could backport this to v6?

Hi @slashmili,

Since there isn't a dedicated maintenance branch for v6 (and releases like v6.0.1 are tagged directly off main), I assume you don't mean cherry-picking for another branch. But if you're asking whether it's okay to release this as a 6.x version, then yes — I believe it can be released as v6.0.2, since it qualifies as a fix (in the conventional-commits sense) and we're currently on v6.0.1.

[5j9]

I meant to version v5!

Oh, sure. I'll create v5 branch and cherry-pick it there.

[5j9]

I've created a new branch dst-v5 based on the v5.3.0 tag (ea119aa) and cherry-picked the fix onto it. If you can create a v5 branch on the upstream repository (here), I'll be able to open a new PR.

[slashmili]

Done! I created v5 branch