refactor(query): decrease call tzdb parse

[TCeason]

I hereby agree to the terms of the CLA available at: https://docs.databend.com/dev/policies/cla/

Summary

• Introduce a timezone lookup-table crate and refactor datetime parsing (cursor_read_datetime_ext.rs) plus arithmetic helpers (date_helper. rs, timestamp scalars) to call into it, covering both UTC→local decomposition and local→UTC
  resolution across DST gaps.
• Update format parsing to cope with jiff 0.2.16 (string_to_format_datetime), reject conflicting %s directives, and leverage the fast path for to_date, add_months, age, etc., while keeping a jiff fallback for out-of-range years.
• Refresh SQL logic suites and the new datetime_fast_path bench so CI can validate DST edge cases and developers can track the observed speed-ups locally.

Benchmarks

• Command: cargo bench -p databend-common-functions --bench bench -- datetime_fast_path
• Results (branch vs upstream main):
• date_add_days: 214.4 µs vs 211 µs (noise-level change)
• string_parse_to_date: 42.16 ms vs 44.43 ms (~5% faster)
• timestamp_add_months: 14.34 ms vs 19.78 ms (~1.4× faster)
• timestamp_extract_components: 12.48 ms vs 20.19 ms (~1.6× faster)

In PR:

cargo bench -p databend-common-functions --bench bench -- datetime_fast_path
    Finished `bench` profile [optimized + debuginfo] target(s) in 0.41s
     Running benches/bench.rs (target/release/deps/bench-dcd523671a1b5430)
Timer precision: 11 ns
bench                               fastest       │ slowest       │ median        │ mean          │ samples │ iters
╰─ datetime_fast_path                             │               │               │               │         │
   ├─ date_add_days                 193.5 µs      │ 353.8 µs      │ 210 µs        │ 214.4 µs      │ 100     │ 100
   ├─ string_parse_to_date          40.03 ms      │ 62.7 ms       │ 40.06 ms      │ 42.16 ms      │ 12      │ 12
   ├─ timestamp_add_months          14.3 ms       │ 14.59 ms      │ 14.33 ms      │ 14.34 ms      │ 35      │ 35
   ╰─ timestamp_extract_components  12.43 ms      │ 12.53 ms      │ 12.47 ms      │ 12.48 ms      │ 41      │ 41

In Main:

cargo bench -p databend-common-functions --bench bench -- datetime_fast_path
    Finished `bench` profile [optimized + debuginfo] target(s) in 0.42s
     Running benches/bench.rs (target/release/deps/bench-caf5866ff0e5233c)
Timer precision: 10 ns
bench                               fastest       │ slowest       │ median        │ mean          │ samples │ iters
╰─ datetime_fast_path                             │               │               │               │         │
   ├─ date_add_days                 193.7 µs      │ 318 µs        │ 210.1 µs      │ 211 µs        │ 100     │ 100
   ├─ string_parse_to_date          44.34 ms      │ 44.63 ms      │ 44.43 ms      │ 44.43 ms      │ 12      │ 12
   ├─ timestamp_add_months          19.7 ms       │ 19.87 ms      │ 19.78 ms      │ 19.78 ms      │ 26      │ 26
   ╰─ timestamp_extract_components  20.16 ms      │ 20.22 ms      │ 20.19 ms      │ 20.19 ms      │ 25      │ 25

Tests

• Unit Test
• Logic Test
• Benchmark Test
• No Test - Explain why

Type of change

• Bug Fix (non-breaking change which fixes an issue)
• New Feature (non-breaking change which adds functionality)
• Breaking Change (fix or feature that could cause existing functionality not to work as expected)
• Documentation Update
• Refactoring
• Performance Improvement
• Other (please describe):

---

This change is 

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".