This is necessary to ensure the implicit scope is consistent when involving match types, since they may or may not have been reduced before implicit search.
We can for example get different results when loading from tasty than when in the same run.

Fixes scala#20071

Make `isMatch` false for applied `MatchAlias`es,
i.e. true only for `MatchType`s and higher-kinded abstraction of them.

As a result, code using `isMatch` to choose between a `TypeAlias` and `MatchAlias`
will now use a `TypeAlias` when aliasing a `MatchAlias`.
Which in turn allows for better de-aliasing, since `dealias` only de-aliases standard type aliases.
The logic for this distinction has also been extracted to the common `AliasingBounds` supertype.

`tryNormalize` on `AppliedType`s should only attempt reduction if there is an underlying match type.
This could previously be identified by a `MatchAlias` tycon. We now need a recursive check.

`def underlyingMatchType` had an `isMatchAlias` guard for `AppliedType`s.
This used to be a quick check preventing unnecessary recursions and superType computations.

But `isMatchAlias` is now itself mutually recursive with `underlyingMatchType`,
so we cache it for AppliedTypes to alleviate this.