[QECGatesCost] Adjustments to match t_complexity rotations, part 2

[mpharrigan]

Following #1313 -- some more changes to how we count.

• there's a QVR stub. The permanent fix would be to use qvr, but removing the erroneous t complexity is a start
• LPRSInterimPrep has a bad t_complexity annotation, based on the current convention for counting CRz as one rotation
• states and effects can be counted (these would throw an error in t_complexity)
• drive by fixes to some str representations so it's clearer what the bloq is