Siqn improvements

[jshipton]

This makes the SIQN timeloop closer to the LFRic timeloop in a couple of very minor ways:

1. The explicit forcing is now applied to xn rather than x_after_slow_phys. Since we haven't yet run anything slow physics this should not change anything.
2. Add the option to offcentre the transporting velocity but default to using a centred transporting velocity. This will only affect any runs that use an offcentring value of alpha that is not equal to a half.

[tommbendall]

Thanks Jemma!