Update bilby settings

After discussions with @aaron.zimmerman, PE co-chair, we almost fixed the sampler settings used at the start of O4, with potential future adjustments. This MR updates bilby settings based on the discussions.

• Set the chirp mass boundary between BNS and potential NSBH to 1.465Msun, computed from component masses of 3Msun and 1Msun, and use low-spin IMRPhenomD (what we called quick_bns) below and high-spin IMRPhenomPv2 (what we called fast_test) above.
• For high mass region, where newly developed IMRPhenomXPHM bases are available, use IMRPhenomXPHM.
• Use these "production" settings by default for MDC and production analyses, and we no longer have multiple bilby runs on a single event. The sample file from the production-mode run is named Bilby.posterior_samples.hdf5 (not Bilby.production.posterior_samples.hdf5), which solves #605 (closed). (I still keep the code structure to enable multiple bilby runs per a single event, as we consider it as a future option.)
• Increase the time limit of condor job to 12 hours, given we reduce the number of runs and we have enough dedicated resources.
• Bump pesummary to 1.0.0.
• Disable ligo skymap and use 6 cpus and online-PE dedicated nodes to speed up the pesummary postprocess.
• Use live data for bilby running on gracedb-test. This just does what !1178 (merged) is expected to do.

These settings are under review (see this review wiki).

cc: @colm.talbot @patricia-schmidt