Avoid interface method dispatches for iterables in Validator

[kyri-petrou]

While looking at some profiles, I noticed a lot of itable stubs in Caliban's Validator. Most of those came from methods such as nonEmpty and toList on classes that don't override them. In addition, in Scala 3 there was a fair bit of overhead when using foreach because it is not inlined (unlike Scala 2).

In this PR, we:

• substitute calls to nonEmpty with either v ne Nil for Lists and !v.isEmpty for other iterables, and in some cases avoid method calls altogether when we know the iterable is empty.
• Add a package-private foreachOne method as a substitute of foreach for lists (which we heavily use in Validator). In Scala 3 this is an inlined while loop, whereas in Scala 2 we simply proxy to foreach (which the compiler inlines)
• Avoid invocation of unapply methods wherever possible especially in cases that we are not using all of the arguments of a case class

This results in a rather noticeably performance improvement (for Scala 3, haven't tested for Scala 2) especially when validating small queries (up to 50% improvement)

series/2.x:

[info] Benchmark                                        Mode  Cnt        Score        Error  Units
[info] ValidationBenchmark.deep                        thrpt   10  1767580.000 ±   8513.908  ops/s
[info] ValidationBenchmark.introspection               thrpt   10   161401.839 ±  11599.050  ops/s
[info] ValidationBenchmark.multifield                  thrpt   10  2143755.893 ±  35898.045  ops/s
[info] ValidationBenchmark.simple                      thrpt   10  4089052.472 ±  60659.943  ops/s

PR:

[info] Benchmark                                        Mode  Cnt        Score        Error  Units
[info] ValidationBenchmark.deep                        thrpt   10  2578415.376 ±  34212.116  ops/s
[info] ValidationBenchmark.introspection               thrpt   10   168605.760 ±   3666.029  ops/s
[info] ValidationBenchmark.multifield                  thrpt   10  2888155.836 ±  12638.780  ops/s
[info] ValidationBenchmark.simple                      thrpt   10  6226946.771 ±  88354.757  ops/s