perf: vectorize permutation test in gravity metric

[cbueth]

I replaced the serial for-loop over permutations with a single rng.permuted call that generates all n_tests permutations as a 2-D array, then computes gravity in a single vectorized expression.

Before serial loop:

count = 0
for _ in range(n_tests):
    np.random.shuffle(b)
    if grav(a, b) > gravity:
        count += 1

After vectorised:

permuted_b = rng.permuted(np.tile(b, (n_tests, 1)), axis=1)
q1 = np.sum(np.exp(a))
q2 = np.sum(np.exp(permuted_b), axis=1)
gravity_permuted = q1 * q2
return np.sum(gravity_permuted > gravity) / n_tests

Speed grows with n_tests, so it would be ~50 times faster at n_tests=10_000.

As gravity is just a toy model, this is not as relevant, but I wanted to contribute this anyways.

[codspeed-hq]

Merging this PR will regress 2 benchmarks

⚠️ Different runtime environments detected

Some benchmarks with significant performance changes were compared across different runtime environments,
which may affect the accuracy of the results.

Open the report in CodSpeed to investigate

⚡ 2 improved benchmarks
❌ 2 regressed benchmarks
✅ 24 untouched benchmarks

Warning

Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
❌	test_detrend_method[z_score-p10]	2.1 ms	3.2 ms	-35.38%
❌	test_ordinal_patterns	15.9 ms	23.5 ms	-32.4%
⚡	test_reconstruct[nodes10-gv]	309.4 ms	112.4 ms	×2.8
⚡	test_reconstruct[nodes5-gv]	67.6 ms	24.6 ms	×2.7

Tip

Investigate this regression by commenting @codspeedbot fix this regression on this PR, or directly use the CodSpeed MCP with your agent.

---

_{Comparing perf/vectorize-gravity (b15c350) 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 (b15c350).

Additional details and impacted files

@@            Coverage Diff            @@
##              main       #54   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           29        29           
  Lines          842       847    +5     
  Branches       151       151           
=========================================
+ Hits           842       847    +5     

🚀 New features to boost your workflow:

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