Implement in-sample predictions in `BATS`/`TBATS` without exceptions

[Mr-Geekman]

🚀 Feature Request

Currently we can't make in-sample predictions with BATS/TBATS, we are throwing exception because underlying library doesn't support it.

However, according to mathematical formulation this possibility exist and we can implement it.

Proposal

Add possibility to make in-sample predictions in BATS/TBATS models without errors.

Implementation from sktime can help (they use the same underlying library).

Test cases

1. Make sure current tests pass.
2. Pay attention to tests in test_inference, after making in-sample possible without the exception we should probably move BATS/TBATS to a new test case.

Additional context

This task waits for #800 to finish.