A new table contains a link to a diff between the current HEAD of master
and the git hash at the time of review.

Comment out action items elsewhere
Create a study page for the impulse response test

- now that LAL is built with pthread locking enabled, and we require it to
  be in the configure script, we can remove what is now redundant fftw
  wisdom protection.  this should enable multi-threaded injection
  generation.
- see PR2228

- was providing hooks for accessing wisdom locking, but no longer required
  because LAL is built with locking enabled

- see PR2220

- see PR2218

- use tools from process support module in glue
- make use of ValueErrors during unpacking to add error checking

- see PR2218

- I don't konw why, but this makes the code run for me.  before this change I was getting tracebacks ending with "ValueError: No JSON object could be decoded", and now I don't get that
- normally I *hate* mystery patches that make problems go away for no apparent reason but this patch reduces the number of gracedb imports by 1 so it's a simplification that has aesthetic value if nothing else, also it's not impossible that by importing the default service URL from one place in the gracedb namespace and then importing the rest api from a different place in the namespace that some sort of conflicting side effects are being triggered related to the multiple generations of legacy apis living together inside the gracedb package.

- see PR2218

- don't mess with the global LIGOLWContentHandler in case it interfers with the guts of other code
- don't import modules and content handler customizations that aren't required
- import utils as ligolw_utils to avoid namespace pollution
- use the .get_table() methods of table classes

- see PR2218

- see PR2218

- see PR2219

- from "strain spectral densities" to "strain spectral density plot"

- if available.

- also change lower bound to 6 Hz from 1 Hz