Scheduling: Calibration expansion does not handle headers correctly

[kalzoo]

User report from the scheduling change made in #336:

• There is an error regarding DECLARE instruction when programs are scheduled, if a DECLARE exists within a DEFCAL. This suggests that the instruction is mishandled during calibration expansion
• There is also an error regarding the PRAGMA instruction timing; this is because there are some zero-time PRAGMA instructions in the standard Rigetti calibrations. The way to handle these should be documented and made easier.