Support handlers that are disabled by their configuration.

[jmalloc]

What change does this introduce?

This PR changes the engine package to skip any disabled handlers by default.

Why make this change?

Further prototyping of the disabling feature before committing to add it to the dogma interfaces.

Is there anything you are unsure about?

• testkit already has its own concept of enabling/disabling handlers. I've opted to disallow explicitly enabling a handler that has been disable in Configure() via the Test interface, but this is allowed via the lower-level interface of the engine package.

• I am not sure what behavior EnableHandlersLike(pattern) should have if it matches a handler that has been explicitly disabled by its configuration. The options are:

  • panic
  • don't enable the handler even though it matches (EDIT: I chose this approach for now)
  • enable the handler anyway (not really a valid option, as it would almost certainly result in some other unexpected behavior)

Ultimately, I'd like to work towards removing the need to enable/disable individual handlers in tests, but I'm not sure that's really feasible.

What issues does this relate to?

• Add Disable() method to configurers. dogma#162
• Automatically enable handlers if they are required for a specific test. #198
• Replace engine options for enabling/disabling handlers with a predicate function. #186

[danilvpetrov]

Looking good to me, except for a commitsh reference (as opposed to a proper tag) to dogma/dogma package in go.mod.