fix: LRL2Classifier honors random_seed config

[immu4989]

Why are these changes needed?

Seed random_state on LRL2Classifier.__init__ for consistency with the rest of the audited estimator family (CatBoost, ElasticNet, LinearSVC, SGD, LRL1, RandomForest, ExtraTrees, all XGBoost variants).

LRL2Classifier's default solver is lbfgs, which is deterministic, so the existing lrl2 reproducibility tests have always passed — but issue #1540 flagged this class as "relevant for sag/saga" solvers. Users who override solver="sag" or solver="saga" via config get a stochastic solver today that doesn't honor the FLAML-internal random_seed. This change closes that edge case using the same defensive pattern shipped in #1541, #1546, #1547, #1549, and #1551.

This is the final estimator in tracking issue #1540. With this merged, every random_state-accepting estimator in flaml/automl/model.py (sklearn family) consistently honors the FLAML random_seed config.

Related issue number

Closes #1540
Pattern reference: #1541 (SGD), #1546 (LRL1), #1547 (RandomForest), #1549 (XGBoost-sklearn), #1551 (XGBoost-native)

Checks

• I've used pre-commit to lint the changes in this PR (note the same in integrated in our CI checks).
• I've included any doc changes needed for https://microsoft.github.io/FLAML/. See https://microsoft.github.io/FLAML/docs/Contribute#documentation to build and test documentation locally.
• I've added tests (if relevant) corresponding to the changes introduced in this PR.
• I've made sure all auto checks have passed.