perf(ordinal-patterns): serial path for small data and bypass decorator overhead

[cbueth]

This PR introduces two changes to the COP (Continuous Ordinal Patterns) connectivity metric:

1. Serial path for small data: norm_windows and pattern_distance used numba parallel=True with prange, spawning OpenMP threads on every call. For small benchmark data (ts_len=50, p_size=3 -> 48 windows), 91% of execution time was OpenMP thread synchronization. Split each into serial (@njit with range) and parallel (@njit with prange) variants, with a plain-Python dispatcher choosing the serial path when n_windows < 200. pattern_transform_2d no longer needs @njit since it delegates to the dispatchers.

2. Decorator bypass: random_patterns called gt_multi_lag (wrapped in @connectivity) inside its per-pattern loop, incurring inspect.signature().bind() overhead for every random pattern. Call find_optimal_lag(gt_single_lag, ...) directly instead. No redundant validation.

Also adds two test types:

• Random-noise tests that verify norm_windows and pattern_distance against a numpy independent reference implementation.
• Serial-vs-parallel equivalence tests confirming both paths produce identical output on >200-window data.

[codspeed-hq]

Merging this PR will improve performance by ×3.5

⚡ 2 improved benchmarks
✅ 26 untouched benchmarks

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
⚡	test_ordinal_patterns	15.9 ms	1.9 ms	×8.1
⚡	test_detrend_method[z_score-p10]	2.1 ms	1.3 ms	+53.09%

Tip

Curious why this is faster? Comment @codspeedbot explain why this is faster on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing perf/ordinal-patterns-serial-path (4858190) with main (875adcd)}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00%. Comparing base (875adcd) to head (4858190).

Additional details and impacted files

@@            Coverage Diff            @@
##              main       #53   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           29        29           
  Lines          842       849    +7     
  Branches       151       152    +1     
=========================================
+ Hits           842       849    +7     

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.