datalog run-time set_cycle isn't robust

[excaliburtb]

after the change to multi-rate logging, the set_cycle method is more complicated in that it must know if it's a pre-init call or post-init so that it can set the next cycle tics according to the current sim tics or for just afterward.

What still works:

dr_group.set_cycle(0.1)
trick.add_read(1.0, 'dr_group.set_cycle(1.0)')
trick.add_read(3.0, 'dr_group.set_cycle(0.1)')

What doesn't:

ss_end = 1800.0
cs_end = 30600.0
cs_logging_break = 28796.6
ph_logging_break = 1796.6

trick.add_read(2.4,"dr_group.set_cycle(ph_logging_break)")
trick.add_read(ss_end-1.1,"dr_group.set_cycle(0.1)")
trick.add_read(ss_end+2.4,"dr_group.set_cycle(cs_logging_break)")
trick.add_read(cs_end-1.1,"dr_group.set_cycle(0.1)")
trick.stop(cs_end+2.6)

i.e.

Setting cycle to 1796.6 at time 2.4
Setting cycle to 0.1 at time 1798.9
Setting cycle to 28796.6 at time 1802.4
Setting cycle to 0.1 at time 30598.9

[excaliburtb]

Update: Actually duplicated the signature for halted logging on the git hash prior to the multi-rate logging update using the following:

ss_end = 1800.0-1789.0
cs_end = 30600.0-20000-1789.0
cs_logging_break = 28796.6-20000-1789.0
ph_logging_break = 1796.6-1789.0

trick.add_read(2.4,"dr_group.set_cycle(ph_logging_break)")
trick.add_read(ss_end-1.1,"dr_group.set_cycle(0.1)")
trick.add_read(ss_end+2.4,"dr_group.set_cycle(cs_logging_break)")
trick.add_read(cs_end-1.1,"dr_group.set_cycle(0.1)")
trick.stop(cs_end+2.6)