Fix SPI initialization to be MSB first

[arlosi]

The comments indicate that the code is attempting to set the SPI order to MSB first, however it actually sets it to LSB first. Since most devices are MSB first, this is the correct choice.

Closes #59.

[arlosi]

With master + this change, I can successfully communicate with two different SPI devices using an STM32F334

[Sh3Rm4n]

Thank you for this PR. Sorry for the delay, I was quite busy with work and uni stuff the last weeks.

It seems like I oversaw this, when I changed these lines. I briefly tested it today and it seems to fix the SPI issue. I will look into this PR this week :)

[Sh3Rm4n]

Okay I've confirmed that your fix is working.

Would be strange, if not 😅

I would be happy, if you added a note to the changelog. Something like:

# Fix

- Fixed regression ... spi was not correctly intialized as master

:)

[Sh3Rm4n]

Thanks!

Meh travis timed out. Force pushing with --amend --no-edit or squashing the commits should retrigger travis.

I should look into fixing travis sometime, as two instances doing the sames is really not useful 😅

[Sh3Rm4n]

Thanks for force pushing. This sadly didn't help. The travis instance has a hard time initializing sscache. I'll look into fixing this.

[Sh3Rm4n]

Disable caching in travis worked 🎉