Add simulation of sync pulse width affecting horizontal position of the image

[richard-broadhurst]

On a CRT, the PLL syncs the horizontal position to the centre of the horizontal sync pulse, beebem does not.
This can be seen in Orlando's 100 scroll demos on the 3D Pool disc or in the demo on https://stardot.org.uk/forums/viewtopic.php?p=234773#p234773.
I think this was only a couple of lines to change in 2.2 but I haven't looked at this in the latest code.
On a CRT, it can take from 2 to 8 scan lines to completely realign in an analogue way, but it is good enough to snap it per scan line.
You can test this on a real beeb with a CRT, b-em, b2, jsbeeb, beebjit, Pi:rgb2hdmi and BeebFPGA.

[chrisn]

Can you share the patch against 2.2? I could figure out how to apply it to the latest code