Fixing weight calculation for covariance of ITk strip cluster on track

Closes ATLSWUPGR-159.

This MR fixes a problem seen in the GX2fitter complaining about measurements with error = 0:

02:03:50 ITkAmbiguitySolver.ITkAmbigui...ITkTrackFitterAmbi     3     2 WARNING Measurement error is zero or negative, drop hit

This is tracked down to be due to a bug in the weight term used to scale the covariance term of clusters on track accordingly to the strip pitch at the estimated track parameter. In rare cases the track parameters return an invalid cellId (outside the sensor) resulting in an invalid evaluation of the strip pitch at the estimated track parameter.

FYI: @nstyles, @pagessin, @tstreble