Don't set status byte on Meta Message (Fix #1574)

[TimFelixBeyer]

Fixes #1574
When loading a MIDI file, messages can set a (running) status byte. However, meta messages don't set the status byte.

This leads to problem on several MIDI files, for example 8584aaa73eb54bae39708da8be5e6282.mid from the lakh dataset.

It could be argued that the != 0xff check is too conservative as running status is only applicable for messages in the range 0x8n - 0xEn, the proposal is this way to be aligned with other python midi libraries:
https://github.com/mido/mido/blob/bff6bf2a8711e33782de5661ed5ccdffb0bb1688/mido/midifiles/midifiles.py#L204-L206

[coveralls]

Coverage: 93.104% (+0.003%) from 93.101% when pulling 2cc61a4 on TimFelixBeyer:patch-4 into 0f9d6cc on cuthbertLab:master.

[mscuthbert]

Only q: Should a meta 0xFF message set lastStatusByte to None instead or is the lastStatusByte still valid?

(btw -- preference for using elif in the patch rather than an else clause followed directly by if with no else)

[TimFelixBeyer]

Only q: Should a meta 0xFF message set lastStatusByte to None instead or is the lastStatusByte still valid?

I believe the lastStatusByte is still valid - the second source I mentioned in #1574 contains this quote about meta events “Also, they can be interleaved within other messages without affecting running status in any way” - that’s how Mido treats it as well.

[mscuthbert]

I believe the lastStatusByte is still valid - the second source I mentioned in #1574 contains this quote about meta events “Also, they can be interleaved within other messages without affecting running status in any way” - that’s how Mido treats it as well.

Great -- thank you for that clarification. With it, I'm thrilled to merge. (The elif is a nice bonus)