fix(trapezoid commutation): non-centred modulation produced no output for Uq < 0

[stijncarelsbergh]

fix(trapezoid commutation): non-centred modulation produced no output for Uq < 0

The trapezoid branches used center = Uq for non-centred modulation, which works
for Uq > 0 but makes all three phase voltages <= 0 for Uq < 0; setPwm() clamps
them to 0, so the motor gets no voltage at all in one direction. The sine/SVPWM
branches handle this by shifting the phases up by their minimum, which is what
non-centred modulation means for every mode.

Using the same min-clamp for the trapezoid modes is exactly equivalent for
Uq > 0 (every sector map contains a -1, so min = -Uq and the result is
map*Uq + Uq as before) and gives the correct shifted waveform for Uq < 0.

---

Split out of #571 at your request: one fix per PR, against dev. The branch contains
nothing else, so it can be reviewed, amended or dropped on its own.

The CI board matrix runs automatically; I did not run any hardware test, so the behavioural
claims are from reading the code plus the compiler. Happy to adjust the wording, split it
differently or drop it - no attachment to this one.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[Candas1]

I once had Claude check the library for bugs and it also mentioned this case.
It's some code I changed so I was surprised.

Nobody probably experienced the bug because it only happens with trapezoidal commutation/non centered/negative torque but it's worth fixing/testing it.

Claude is telling me there is a simpler and equivalent fix:
center = modulation_centered ? (driver->voltage_limit)/2 : fabsf(Uq);