Closes #1496

Changes proposed in this Pull Request

When using transformer types in stochastic networks, apply_transformer_types() failed with because it was not handling multiindex with scenarios when matching data with transformer_types table.

Fix is to drop scenarios & look at unique transformer types. User changes across type data are caught by the consistency checks.

Small history on this: when preparing #1154 we did think about problem of types broadcasting, and had exactly this fix implemented, which was reverted before merge since we universally agreed types should ideally not be broadcasted with scenarios (w/o re-applying the fix). Problems of this kind will go with #1474

@gincrement thanks for putting attention to this. One question, in your version you've included "tap_ratio" into a copy of transformers data: https://github.com/PyPSA/PyPSA/compare/master...gincrement:pypsa:patch-9?diff=split&w#diff-4d3697f16fbe48a31c279ea3a292649b3c6a62aebe09e837b34c35674af6e30bR484. I don't see a reason for doing this, since tap_ratio is created fresh later in function and not extracted from either table. Do I miss something?

Checklist

• Code changes are sufficiently documented; i.e. new functions contain docstrings and further explanations may be given in docs.
• Unit tests for new features were added (if applicable).
• A note for the release notes docs/release-notes.md of the upcoming release is included.
• I consent to the release of this PR's code under the MIT license.