Fix DST timezone handling for viewed event weeks

[wslany]

Fixes #227

This PR is intentionally split into several commits:

1. 4ceda5a adds the DST regression proof, focused helper tests, and frontend CI.
2. ece4641 applies the actual fix that makes the remaining regression pass.
3. 5b76640 cleanup that removes a deprecated Vite CJS warning
4. 55dd935 adds another regression test showing the correctness for fall-back DST changes, and another test for the non-edge case where the standard offset is kept during non-DST viewed weeks

I split it this way so review can follow a clear red/green sequence:

• first, prove the bug exists and that the regression test catches it,
• then, apply the smallest fix that turns the remaining failing regression green.

The first commit also includes unit tests for the new helper functions so the extracted date/offset logic is documented directly in tests, not only in implementation.

[wslany]

Red proof for commit 4ceda5a.

At this point the branch contains:

• the failing DST regression test,
• focused helper tests for the new date/offset helpers,
• frontend CI wiring.

So the first commit is intentionally structured with:

• helper tests green,
• the user-facing DST regression still red.

Local result at commit:
npm run test:unit

• 4 passed
• 1 failed

Failing case:
shows the viewed event week using that week's timezone offset

That failing result is intentional here: it demonstrates the bug exists and that the regression test really catches it before the fix is applied.

[wslany]

Green proof for commit ece4641.

This second commit wires the scheduling view and timezone selector to the new reference-date helpers so both the scheduling math and the visible GMT label are derived from the viewed event week instead of the current machine date.

Local result at commit:
npm run test:unit

• 5 passed

I also verified:
npm run build

• build succeeds

So the red/green sequence is now:

• commit 1: helper tests green, DST regression red
• commit 2: all tests green after the actual fix

[wslany]

I added one small follow-up cleanup commit: vitest.config.js is now vitest.config.mjs, which removes the deprecated Vite CJS warning from the test output without changing behavior.

[wslany]

I added one more small follow-up test commit so the viewed-week timezone tests now cover:

• a normal non-DST case,
• the spring-forward transition,
• the fall-back transition.

That gives a bit more confidence that the fix is not just handling one DST edge case, but the general viewed-week offset behavior.

[maxwellward]

Looks good to me!

[wslany]

@wslany wslany requested a review from maxwellward

Sorry, my mistake, please ignore the re-request. And many thanks for the feedback!