Add a PermissiveRespondTo extension for Mashes

[michaelherold]

By default, Mashes don't state that they respond to unset keys. This
causes unexpected behavior when you try to use a Mash with a
SimpleDelegator.

This new extension allows you create a permissive subclass of Mash that
will be fully compatible with SimpleDelegator and allow you to fully do
thunk-oriented programming with Mashes.

This comes with the trade-off of a ~19KB cache for each of these
subclasses and a ~20% performance penalty on any of those subclasses.

Closes #461

[dblock]

@BobbyMcWho want to review this?

[BobbyMcWho]

This comes at the cost of approximately 20% performance for initialization and setters and 19KB of permanent memory growth for each such class that you create.

😬 Is this per class, or per instance?

Regardless, this looks useful in niche cases where someone really wants to use a Mash, I'd argue they should probably use a Dash unless they're just delegating methods that they don't have insight into whether they'd exist on the Mash.

LGTM

[michaelherold]

Is this per class, or per instance?

The 19KB is per class. It's because each time you do include Hashie::Extensions::Mash::PermissiveRespondTo, we create a method cache on the singleton class of the includer. That's there to make sure we only have a 20% performance penalty instead of an order-of-magnitude penalty.

The performance penalty is only on the including class.

[dblock]

I am good with this, @BobbyMcWho merge at will.