Add extra test file SwapInst3.it:
note + inst + porta with same inst as before
but with note + inst mapping to another sample
-> must still use old sample (ex: "In the Forge.it")

See notes in StrongHelp file for limitations of each file format.

Some effects behave differently on various trackers.
So ideally we need to check for all trackers behaviours like:
- does the effect apply on frame-0, non-0 frames, all frames
  or weirder yet (see MDL retrigs)
- unit of vibrato versus normal pitch slides (usually vibrato is 2x slide)
- unit of tremolo versus normal volume slides
- unit of panbrello versus normal panning slides
- speed of vibrato/tremolo/panbrello may differ
- pitch limits (often larger than note range)
- retrig, tremor bahaviour
- sharing of "memory" between effects (slide up.down may have separate mem.)
- note delay of 0
- note cut with delay of 0
- sample offset, finetune: do they work anytime or only for new notes?
- do effects in note column ("key off", "note cut") affect last note memory?
- behaviour of <> combinations of note + inst + portamento
- behaviour of new inst in inst column with inst > max inst, non-stored sample,
  sample with len of (cuts current note? applies duplicates note check
  on virtual notes? forces volume to default sample volume? ...)

Sample offset slides:
DBM 18xx -> slide
MMD 20xx -> add
GT2 BCxx -> add
    BDxx -> remove

Digi/DBM:
E3x also without note?

DSKT:
06xx -> play without loop from offset
08xx -> add to sample
09xx -> add to sample x256
0Exy -> slide by Y ??? X (panning)

IT:
T00 may reuse last slide or tempo
Envelopes: when turned off "current envelope value" is still taken into account
instead of bypassed.

Backward play/reverse combined with set sample offset: offset is to be used
from start or from end of sample?

Multiple SEx on a row, test indicates first one is used, doc the last one

MT2:
IT cutoff values read as: 130, 138, 142, ..., 4698, 5123 Hz
so does seem to differ from MT2.
Channel filters(delay, reverb, ...)/equalisers are not implemented.
Most automatons are not implemented.
High pass filter not implemented.
Filer attack/decay not implemented (seems to be used only if no envelope)
Synchronlised samples are not implemented

PTM:
note slide + retrig, implementation is wrong (Abnormality.ptm)

ULT:
Combined Reverse + Offset, must we reverse offset?
Keyoff does not reset note nr.
effect 50c ->sustain off???

XM:
new note, if semitone + reltone outsside [C-0, B-9[ old note continues to play
F00 stops tracker