fix: `grind` introduces the hypotheses of a goal wrapped in `mdata`

This PR fixes `grind` not introducing the hypotheses of an implication or universal quantifier when the goal type is wrapped in `Expr.mdata`, as it is after a `have` (`noImplicitLambda`). The goal was negated as a whole instead, so a hypothesis `¬(p → q)` reached the normalizer and the solvers.

`initCore` now erases the irrelevant metadata of the target, with the existing `eraseIrrelevantMData`, before the introduction loop runs.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
4 files changed