Speed optimisations of PixelDiodeMatrix

Additional speed optimisations of PixelDiodeMatrix:

• precompute 1/pitch and use that for calculations
• reuse bool outOfBounds = index >= m_numCells; calculation
• do not check index < 0 but rather index > 0 and only do actions when needed
• (also fixed some whitespace)

This yields about 40% faster execution. There may be some numerical differences at the end. Let's see what the CI will say.

/cc @nstyles @jchapman @tdado