Add a simple profiling system to liquid rendering. Each
liquid tag ({{ }} and {% %}) is processed through this profiling,
keeping track of the partial name (in the case of {% include %}), line
number, and the time it took to render the tag. In the case of {%
include %}, the profiler keeps track of the name of the partial and
properly links back tag rendering to the partial and line number for
easy lookup and dive down. With this, it's now possible to track down
exactly how long each tag takes to render.
These hooks get installed and uninstalled on an as-need basis so by
default there is no impact on the overall liquid execution speed.
After moving the method existence check from Context into Strainer,
updated Strainer to only accept invokation methods that were added via
filter Modules, and done in a way that respond_to? is never called,
preventing unconstrained Symbol table growth.
Class.public_method_defined? ends up diving into Ruby's core looking for
a method with the given method_or_key. This process at some point turns
method_or_key into a Symbol. This change no longer takes that path and
thus doesn't grow the Symbol table.