np2: experimental feature-complete SSG-EG

[jpcima]

Here is full SSG-EG including support: ALT/ATT/HOLD

I took Nuked OPN2 as basis for comparison.
I plotted the different envelopes and fixed until it matches approximately.
(16 different tests: 8 ssg modes × 2 test situations)

There was 2 cases:

• first is the ordinary case where AR=$1f. no issue
• second is the "undefined behavior" case AR≠$1f, according to documents.
  In this special case, "alternate" mode toggles "inversion" bit at every cycle of attack. As such, it seems difficult to determine the final polarity with certainty, other than compute with cycle accuracy.
  I made a guess matching with a test instrument, and implemented for the case AR≠$1f. It's probably wrong in cases. As a general rule, SSG-EG attack is best set to zero time.
  "Alternance" flag will produce an initial glitch when there is attack time. I did not replicate this in NP2.