2 Point calibration fix

[redoxcode]

This PR should fix an oversight in the way the calibration values are applied when using only 2 Points.
The issue was described here: #221 (comment)
In short: If we are not mistaken the adc offset measured when 0V is applied, is currently not taken into account.
Old:
result = 'adc raw value' / 'calibration at 10V'
New:
result = ('adc raw value' - 'calibration at 0V') / ('calibration at 10V' - 'calibration at 0V)

(some tests still fail, but it's only one contrib script (tests/contrib/test_europi_turing_machine.py) so I guess it's an issue of this test and not the code changes)

[awonak]

@redoxcode do you have some time to sync this branch to HEAD, and fix the lint and test errors? We'd love to include this in our next release.

[redoxcode]

@awonak sync is done. I will try to work out why tests fail.
the lint part shouldn't be a problem.

[redoxcode]

ok, there where some bugs in the implementation and some corner cases I missed (like when the input is below the 0 point calibration). But I think its good now!
But please check on actual hardware and do the calibration.
I also had to fix the 2 point calibration in the mock hardware.
And some how my black output was different than what lint wants 🤔

[mjaskula]

@redoxcode Thanks for taking the time to dig into this, and thanks to @francoisgeorgy for describing the issue and solution in the first place. All of these changes look good to me. This should help some of our 0v reading issues.

Also, selfishly, I'm glad that the test failures didn't end up being issues in the turing machine tests themselves.

[awonak]

LGTM, thank you for improving the calibration process @redoxcode & @francoisgeorgy