perf: Add fastpath for when rounding by single constant durations

[MarcoGorelli]

Combines:

• perf: Make dt.truncate 1.5x faster when every is just a single duration (and not an expression) #16666
• perf: make truncate 4x faster in simple cases #16615

and implements them for round. I've included a hypothesis test right away to check this doesn't introduce regressions

I'll pick https://github.com/pola-rs/polars/pull/12597/files up again after this one, let's try to get it across the line

---

drive-by: moving some .dt.round tests to test_round.py

[codecov]

Codecov Report

Attention: Patch coverage is 94.80519% with 4 lines in your changes missing coverage. Please review.

Project coverage is 80.45%. Comparing base (b33d3f1) to head (3a2b84f).
Report is 13 commits behind head on main.

Files	Patch %	Lines
crates/polars-time/src/round.rs	94.80%	4 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #17580      +/-   ##
==========================================
- Coverage   80.45%   80.45%   -0.01%     
==========================================
  Files        1484     1484              
  Lines      195324   195388      +64     
  Branches     2778     2778              
==========================================
+ Hits       157149   157191      +42     
- Misses      37663    37685      +22     
  Partials      512      512              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.