Optimize _blank?

[moberegger]

Switches _blank? from BLANK == value to Blank === value.

Originally _blank? did a Ruby-level dispatch to Blank#==, which itself calls super (for an identity check) and then Blank === other. We can short circuit this to simply be Blank === other. This gives us the same result for less work because Blank#== would always return true for any Blank instance. We don't actually need to an identity check.

blank1 = Jbuilder::Blank.new
blank2 = Jbuilder::Blank.new

puts blank1 == blank2 # true
puts blank2 == blank1 # true

Below is an IPS benchmark against a simple builder that exercises _blank a few times.

json.foo do
  json.bar 123
end

ruby 4.0.6 (2026-07-14 revision 03b6d3f889) +PRISM [arm64-darwin23]
Warming up --------------------------------------
                 old    50.799k i/100ms
                 new    61.777k i/100ms
Calculating -------------------------------------
                 old    500.692k (± 2.6%) i/s    (2.00 μs/i) -      2.540M in   5.072875s
                 new    643.490k (± 1.4%) i/s    (1.55 μs/i) -      3.274M in   5.088164s

Comparison:
old:   500692.4 i/s
new:   643489.7 i/s - 1.29x  faster

[moberegger]

Not necessary for the optimization, but figured I'd remove it since it is no longer running in the gem.

I suppose it's possible the a user is doing something with Jbuilder::Blank directly that may require this, but I believe this class was intended to be an implementation detail.