Add isExpr, asExpr, asExprOf, asTerm, asTypeTree and asQuotedType

[nicolasstucki]

These replace the Expr methods seal, sealOpt and unseal with equivalent methods with more intuitive names.
Instead of sealOpt we provide a more flexible isExpr. All these methods are defined directly on Tree
instead of Term to avoid the need to perform type test to a Term first.

We similarly replace quoted.Type.seal with asTypeTree and reflect.Type.seal with asQuotedType.

Note: #8940 would remove the asTerm and asTypeTree as there we have Expr[T] <:< Term and something similar for Type, but the others would still be relevant.

[liufengyun]

The names seal and unseal look good to me, meta-programmers get to know them better in usage.

That said, I'm open to changes, the only worry is about compatibility -- it seems the change is pervasive and will break most macros written for Dotty.

[nicolasstucki]

The old version will still be supported for some time with a deprecation warning

[nicolasstucki]

The names seal and unseal also look good to me because I have used them for years. But not to a newcomer that wants to convert an Expr to a Term, Term to an Expr, and so on ... The new names tell exactly what is the conversion and can be easily found by autocompletion.

The toExprOf is the most important one as it gives a way to get an Expr[T] from any Expr or Tree while checking that it is an expression and that it is in fact of type T. I saw too many instances of term.seal.cast[T] and term.seal.asInstanceOf[Expr[T]]; the second one is particularly dangerous as the it is an unchecked cast that leads to unsoundness that shows in -Ycheck in later phases.

[AdamPaynter]

I can confirm that, as a very new newcomer, this makes more sense. In fact, I hadn't even figured out what seal and unseal were until I read this PR. Even now, I still haven't grokked the relationship between Term and Expr (and why you would switch between them). Time for some more reading...

[nicolasstucki]

Replaced with #9514. Will do this in smaller parts