[BUG] Duplicated `_random_ss_ix` helper function across `bootstrap.py`, `enbpi.py`, and `_bagging.py`

[krsatyamthakur-droid]

Describe the issue

The internal helper function _random_ss_ix (randomly uniformly sample indices from a list of indices) is duplicated across three files in the regression module:

1. skpro/regression/bootstrap.py (lines 208–215)
2. skpro/regression/enbpi.py (lines 343–350)
3. skpro/regression/ensemble/_bagging.py (lines 253–257)

The bootstrap.py and enbpi.py versions are identical (with a random_state parameter), while the _bagging.py version is a simplified variant (no random_state parameter, uses np.random.choice directly on the global state).

This duplication increases maintenance burden — any bug fix or enhancement must be applied in all three places.

Suggested fix

Extract a single canonical _random_ss_ix function into skpro/utils/sampling.py (or another appropriate shared utility module), and update all three files to import from there.

The consolidated version should support the random_state parameter (as in bootstrap.py / enbpi.py), making the _bagging.py version also properly support local random state.

Files affected

• skpro/regression/bootstrap.py — delete local _random_ss_ix, import from utils
• skpro/regression/enbpi.py — delete local _random_ss_ix, import from utils
• skpro/regression/ensemble/_bagging.py — delete local _random_ss_ix, import from utils
• skpro/utils/sampling.py — [NEW] shared utility module

Expected impact

• Reduced code duplication
• Easier maintenance — single point of truth for the sampling logic
• Consistent random state handling across all three estimators

I am happy to submit a PR for this.

[fkiraly]

good idea!