Fix flaky genTinySurplusTx by computing surplus dynamically

[Jimbo4350]

Changelog

- description: |
    Fix flaky genTinySurplusTx test by computing surplus relative to actual fee
  type:
   - test
  projects:
   - cardano-api

Context

The genTinySurplusTx generator used a hardcoded surplus range (237–250), assuming the 1-output fee (F1) was always ~236. This was flaky because F1 varies with the randomly generated address structure (different credential types and staking references produce different CBOR serialization sizes). A surplus of 237 was observed to balance successfully on Windows GHC 9.12 CI (failure on PR #1156).

The fix computes F1 per-test via calcMinFeeTx on the generated tx and sets surplus to [F1+4, F1+10] — always in the danger zone regardless of address size.

Also:

• Removes duplicate genTinySurplusTx and property test from Experimental.hs (now lives only in Fee.hs)
• Adds exampleProtocolParamsEra :: Era era -> PParams (LedgerEra era) to keep the generator polymorphic

How to trust this PR

cabal test cardano-api-test --test-option='--pattern=Tiny surplus' --test-option='--hedgehog-tests=5000'

Checklist

• Commit sequence broadly makes sense and commits have useful messages
• New tests are added if needed and existing tests are updated
• Self-reviewed the diff

[Copilot]

Pull request overview

Tightens the genTinySurplusTx test generator’s surplus range to avoid values near the F2 fee threshold and eliminate flakiness in fee-related tests.

Changes:

• Reduce the generated surplus upper bound from 250 to 246 to stay clearly below F2.
• Update generator comments to document the F1/F2 relationship and boundary-case rationale.