Again removing much more. This time with more focus on what the ESS
would not need. Also gives faster development cycle for conda package
development"

Also contains some hacky approaches to excluding test targets

This reduces the payload of Framework. Caveats:
* Implementation is not idea. Firstly, libaries should be managed by
list which would make it easier to harmoise enable sub-directories and
libraries making up the Framework target. Also in the top level cmake,
the enabling and checking could be refactored to a common function given
that it is reasonably generic
* I am probably throwing out more than would be accepted

Changes here are temporary (mainly hacks) to see what can be done about
getting round the assumption that if you are running the install target,
you are also trying to build either MantidPlot or MantidWorkbench

Re #23012

Re #23012

This no longer works and has been unused for some time.
It is superceded by the ENABLE_MANTIDPLOT & ENABLE_WORKBENCH
flags.

With openmp enabled access to the ConfigService is now required

Incorrect options were used to compile the Fortran such
that it produced incorrect results.
Refs #27744

Refs #16578

Dependencies now go into Frameworks as seems to be done for
other application bundles. This should bring us close to
be able to sign the app bundle

- Makes the `FrameworkTests` dependent on `Framework`
- Builds `mantidqt_iconsqt4` when `mantidqt` is built

After finding [this
article](https://stackoverflow.com/questions/36358217/what-is-the-difference-between-option-and-set-cache-bool-for-a-cmake-variabl),
clean up the cmake configuration to use `option`.

Co-Authored-By: alicerussell1 <46603316+alicerussell1@users.noreply.github.com>

This library has gone unused for a number of years.

(May re-instate old ICAT code as a second, backup searcher, and also need to create one for HFIR instruments with run numbers.)

This fixes #21360

Settings for MessageDisplay are not persisted yet.
Refs #21251