Optimize effect transforms

[adroitwhiz]

Proposed Changes

• Replace Drawable.getEnabledEffects() function with Drawable.enabledEffects
• Only apply effect transform in Drawable.getLocalPosition and Drawable.sampleColor4b if drawable effects are enabled
• Replace possiblyDefinedArgument || defaultValue pattern with if (typeof possiblyDefinedArgument === 'undefined') possiblyDefinedArgument = defaultValue in EffectTransform methods

Reason for Changes

• Replace Drawable.getEnabledEffects() function with Drawable.enabledEffects
  • The EffectTransform.transform[Color/Point] methods, called once per pixel in touching tests, get a Drawable's enabled effects. Replacing the pseudo-getter function with a property speeds it up.
• Only apply effect transform in Drawable.getLocalPosition and Drawable.sampleColor4b if drawable effects are enabled
  • This is what I did in Return early from EffectTransform methods if no effect bits are set #481, but taken one step back to the functions that call the effect transform methods in question. This alone provides a bigger performance boost than Return early from EffectTransform methods if no effect bits are set #481.
• Replace possiblyDefinedArgument || defaultValue pattern with if (typeof possiblyDefinedArgument === 'undefined') possiblyDefinedArgument = defaultValue in EffectTransform methods
  • This provides a small performance boost, possibly by removing the "truthy/falsy cast".

This gives a ~40% performance improvement to "touching color" checks on drawables with no effects enabled, and a ~15% performance improvement to ones with effects enabled, as seen in this benchmark:

Current	This PR

[adroitwhiz]

this will conflict with https://github.com/LLK/scratch-render/pull/479/files#diff-3dbe88c4ff1ea0a8c0b27759902040f5R44

[cwillisf]

Thanks for simplifying this stuff! I have one suggestion for further simplification but let me know if you can think of a reason not to do so.

[cwillisf]

👍 thanks!