Add auto-calibrating to most decoders

[crankyoldgit]

Highlights:

• Define timings for protocols where possible, in terms of the greatest common factor.
  i.e. various times are expressed as a multiple of a common tick/period value.
• Use the incoming header mark (& space) to adjust timings for the rest of a protocol's decode routine.
• Phase in use of a common data decoding routine.
• Adjust unit-tests to new timings when they have been redefined due to the 'tick' calculations.
• Unit-test case for a out-of-skew NEC-like message from possibly odd sending device (Apple TV remote Inconsistent decoding for unknown. #264 )

This PR improves matching of messages from sources that are slightly askew from the standard.
In short, it tries to make the timings relative to itself, rather than rely solely on fixed values with a percentage variance.
This addresses a problem discovered in #264 where an Apple TV remote was producing a NEC-like message (fixed in a different PR) but had timings that should have had the NEC decoder work, but it wasn't because the 'space' values were slightly to large. (Out of spec range).
Given the existing code matches that odd example's headers, we can use that to determine that the device is a bit out of spec, and thus adjust to match the message's skew.

[crankyoldgit]

Friendly ping for reviewers.

[markszabo]

LGTM