This is particularly required to be able to use this algorithm in the
post-processing step of a live data run. Unless this new property is
set to false, things fail on the second time around because the monitor
workspace has been cleared out of the accumulation workspace.

This is in addition to the existing way of an additional ADS entry with
"_monitors" suffix. It allows clients to use ExtractMonitorWorkspace on
the main output workspace and not need to worry about the name of the
workspace (it will work if LoadPreNexus was run as a subalgorithm, which
is not the case for old way).

Refs #9599

This is in addition to the existing ways of adding the workspace
separately to the ADS with a "_monitors" suffix and returning
it as an output property. This will allow the new way to be used going
forward without breaking the old way.

The converted text is stored in separate .rst files in the docs directory.
Refs #9562

Refs #9523

Also, the CanSAS format expects "A" rather than "Angstrom".

It looks like history was loaded here at some point in the past. May as well remove this
while I'm here.

SaveNexus and SaveNexusProcessed wouldn't append a record to the history.
This commit fixes that issue and leaves a record on the input workspace so the result is that we get a
complete history, but we lose the execution start time and duration.

Refs #9435