Add cached, named Visibility profiles

[rmosolgo]

This aims to add to improvements to GraphQL-Ruby's (forthcoming) schema visibility system.

• Ahead-of-time caching of named profiles, so that they can be reused verbatim at runtime (no calls to .visible? during execution)
• Along with dynamic: true, which is how the current visibility system always works, calling .visible? for every schema member on every query.

Part of #5014

TODO

• Preload named profiles when preload: true (defaults to true when Rails.env.production?)
• Accept configured visible_in: ... or similar, for named visibility profiles
  • This is still possible, but I don't think it's pressing
• Rename Visibility::Subset => Visibility::Profile
• Document Visibility::Profile

[rmosolgo]

To benchmark this, I modified profile_large_result:

    schema = Class.new(ProfileLargeResult::Schema) do
      # def self.visible?(member, context)
      #   sleep 0.001
      #   super
      # end
    end

    schema2 = Class.new(schema) {
      use GraphQL::Schema::Visibility
    }
    document = ProfileLargeResult::ALL_FIELDS
    Benchmark.ips do |x|
      x.config(time: 10)
      x.report("[WARDEN] Querying for #{ProfileLargeResult::DATA.size} objects") {
        schema.execute(document: document)
      }
      x.report("[PROFILE] Querying for #{ProfileLargeResult::DATA.size} objects") {
        schema2.execute(document: document)
      }

The performance was about the same between the two (Warden a bit faster? within the margin of error.), except when a cached visibility profile was used and sleep 0.001 was present -- then Visibility::Profile was much faster.

Visibility::Profile uses a hair less memory (less than 1% difference).