Composable model

[kdund]

At the moment, the likelihood composability happens inside the statistical model-- but ideally we'd be able to do so also by adding together multiple statistical models as long as they're statistically independent and don't have conflicting parameter demands.

This PR draft includes a very naive start.

from alea.examples import gaussian_model
from alea.model import StatisticalModel, CompoundStatisticalModel
model_a = gaussian_model.GaussianModel()
model_b = gaussian_model.GaussianModel()

cm = CompoundStatisticalModel([model_a, model_b])

cm.data = cm.generate_data(mu=0, sigma=1)

cm.parameters.mu.nominal_value = 0.

cm.fit( sigma=1)

[coveralls]

Pull Request Test Coverage Report for Build 8823038063

Details

• 56 of 129 (43.41%) changed or added relevant lines in 3 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage decreased (-3.6%) to 88.367%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
alea/submitter.py	14	22	63.64%
alea/model.py	13	42	30.95%
alea/runner.py	29	65	44.62%

Totals
Change from base Build 8706621084:	-3.6%
Covered Lines:	1580
Relevant Lines:	1788

---

💛 - Coveralls

[kdund]

Things required before review:

[] Constructor from configs
[] Per-likelihood config overrides
[] compatible with runner
[] compatible with batch submission
[] testing

[dachengx]

@kdund I tried to fix the error in test_init_signatures : https://github.com/XENONnT/alea/actions/runs/8706626297/job/23879636821?pr=154#step:6:201.