1.5.5 fix for critical regression in wire and regression in pinConfigure()

Critical Bugfix Wire "master and slave" mode malfunctioned in 1.5.x versions. This is corrected.

Bugfix USERSIG library never worked correctly. This is corrected.

Bugfix pinConfigure functionality had major regression and failed to work under many circumstances due to a typo, that is corrected.

Bugfix Eliminate use of RAMPZ in attach ISR on parts with 64k flash or less, saving 4 instruction words and 5 clocks.

Bugfix Correct Azduino board bootloader names and options.

Enhancement Experimentation has found that some E-spec 48 MHz parts actually run as "expected" , if your expectations tend towards optimism.... Yeah, it seemed to work fine shrug So the core now has that as an option (for crystal. not just clock)

Enhancement The optimization level menu was brought over from mTC. In addition to letting you choose -O3, it's been found that a large but unpredictable impact on binary size can be realized by turning off GCSE. Whether this makes the binary larger or smaller is hard to predict, but the effect can be over 5%.

Bugfix Issues with converting between analog and digital pins on some DD-series parts resolved. Optiboot DD28's were improperly using the 32-pin variant, too, and an assortment of parts was using wrong settings due to copy-paste errors