Fix DMA starving SPI

[bugadani]

Fixes #2151

Thank you, Dominic

[JurajSadel]

I tried to run spi_loopback_dma on C6 and I'm getting this:

I (125) boot: Disabling RNG early entropy source...
[0, 1, 2, 3, 4, 5, 6, 7, 8, 9] .. [73, 74, 75, 76, 77, 78, 79, 7a, 7b, 0]
[1, 0, 0, 0, 1, 1, 1, 1, 1, 1] .. [1, 1, 1, 1, 1, 1, 1, 1, 1, 1]
[2, 0, 0, 0, 1, 1, 1, 1, 1, 1] .. [1, 1, 1, 1, 1, 1, 1, 1, 1, 1]
[3, 0, 0, 0, 1, 1, 1, 1, 1, 1] .. [1, 1, 1, 1, 1, 1, 1, 1, 1, 1]

[bugadani]

Oh wow you're right. Previously even my first iteration was bad so I assumed that fixing the test was enough, but alas, not.

[Dominaezzz]

Will it still work if you skip the first call to self.update()?

[bugadani]

You mean line 2234? Apparently it does, yes. Funnily enough we had a start_operation that did the update for us, not sure why it wasn't used :D

[Dominaezzz]

Yeah that one. Nice! At least that matches esp-idf's code, which works.

[bjoernQ]

LGTM - great we have this tested now

[JurajSadel]

Thanks!