Add multiprocessing capability to independent sampler

[TomWagg]

I noticed that sampling was getting a bit slow once the number of binaries exceeded 10,000. I realised that COSMIC was doing sampling single-threaded. I don't think we need to do this?

So I've added options for pool and nproc (similar to Evolve) that can be used to parallelise the sampling. For a sample of 1 million binaries I can get a ~14x speedup when using 16 cores on my laptop (2.2s instead of 32s), I think this should approximately scale with the number of cores for large enough samples.

For now I set the minimum chunk size to 1000 binaries so that people don't overdo it on splitting up the sampling.

(Schwimmbad confuses me so feel free to swap that in for the basic multiprocesing pool I used @katiebreivik)

[codecov]

Codecov Report

❌ Patch coverage is 0% with 28 lines in your changes missing coverage. Please review.
✅ Project coverage is 77.65%. Comparing base (8772c07) to head (bbb763d).
⚠️ Report is 143 commits behind head on develop.

Files with missing lines	Patch %	Lines
src/cosmic/sample/sampler/independent.py	0.00%	27 Missing ⚠️
src/cosmic/_version.py	0.00%	1 Missing ⚠️

❌ Your patch status has failed because the patch coverage (0.00%) is below the target coverage (90.00%). You can increase the patch coverage or adjust the target coverage.

Additional details and impacted files

@@             Coverage Diff             @@
##           develop     #771      +/-   ##
===========================================
- Coverage    86.91%   77.65%   -9.26%     
===========================================
  Files           40       50      +10     
  Lines        25542    27579    +2037     
  Branches         0      875     +875     
===========================================
- Hits         22198    21414     -784     
- Misses        3344     5878    +2534     
- Partials         0      287     +287     

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

🚀 New features to boost your workflow:

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

[TomWagg]

You can test the speedups with this code :)

from cosmic.sample.initialbinarytable import InitialBinaryTable
from time import time
import numpy as np

final_kstar = [13, 14]

sampling_params = {
    'qmin': 0.0,
    'm2_min': 0.08,
    "binfrac_model": 1.0,
    "SF_start": 12000,
    "SF_duration": 0.0,
    "met": 0.02,
    "sampling_target": "size",
    "trim_extra_samples": False,
    "q_power_law": 0.0,
    "primary_model": "kroupa01",
    "ecc_model": "sana12",
    "porb_model": "sana12"
}

print("Sampling speed test time!!")
print("--------------------------\n")

for size in [10_000, 100_000, 1_000_000]:
    for nproc in [1, 4, 16, 32]:
        print(f"Sampling {size} binaries with nproc={nproc}...")

        start = time()
        ibt, mass_singles, mass_binaries, n_singles_req, n_bin_req = InitialBinaryTable.sampler(
            'independent', final_kstar, final_kstar, size=size, **sampling_params, nproc=nproc
        )
        end = time()
        print(f"    Took {end - start:.2f} seconds")
        print("    End results:")
        print(f"        Number of binaries sampled: {len(ibt)}")
        print(f"        Mass in singles: {mass_singles:.2f} Msun")
        print(f"        Mass in binaries: {mass_binaries:.2f} Msun")
        print(f"        Number of singles required: {n_singles_req}")
        print(f"        Number of binaries required: {n_bin_req}")
        print(f"        Primary mass percentiles: {np.percentile(ibt['mass_1'], [2.5, 50, 97.5])}")
    print()

[katiebreivik]

I think it's totally fine to just use the regular multiprocessing here. Thanks @TomWagg !