Micro-optimization to AmazingPrint::ActiveRecord#cast

[jdelStrother]

I kept seeing Module#ancestors show up in some profiling involving SemanticLogger.

Changing to x.is_a?(ActiveRecord::Relation) and y <= ActiveRecord::Base should be equivalent, but makes AmazingPrint::Inspector#awesome 7% faster:

require "rails"
require "active_record"
require "amazing_print"
require "benchmark/ips"

inspector = AmazingPrint::Inspector.new
log = { foo: { bar: ["baz"] } }
Benchmark.ips do |x|
  x.report("ancestors.include?") do |times|
    i = 0
    $ap_ancestors = true
    while i < times
      inspector.awesome(log)
      i += 1
    end
  end
  x.report("class <= ") do |times|
    i = 0
    $ap_ancestors = false
    while i < times
      inspector.awesome(log)
      i += 1
    end
  end

  x.compare!
end

ruby 3.2.4 (2024-04-23 revision af471c0e01) [arm64-darwin23]
Warming up --------------------------------------
  ancestors.include?     4.081k i/100ms
           class <=      4.289k i/100ms
Calculating -------------------------------------
  ancestors.include?     40.419k (± 1.5%) i/s -    204.050k in   5.049537s
           class <=      43.345k (± 1.5%) i/s -    218.739k in   5.047616s

Comparison:
           class <= :    43344.7 i/s
  ancestors.include?:    40418.6 i/s - 1.07x  slower

I might be missing something though - any idea why ancestors was used in the first place?