Task #10339: try import, skip test if package not available. package mock is required for build, just for run.

Task #10339: split MoMQueryService into client and server part. Made LTAIngestAdminServer only dependent on MoMQueryServiceClient so we don't run an extra momqueryservice on the lexar3/4

Task #10368: fabric is not required during build, but during runtime. So try to import, and explain what to do when fabric is not installed. The build-host is also not necessarily the run-host

Task #10339: it's not exactly clear what happened to the 2_20 branch, but many changes in LTAIngest from 2_19 were lost. Did a manual merge from 2_19 branch into this 2_20 branch for LTAIngest and SAS/DataManagement

Task #10339: applied patch from 2.19: when setting status and message, set status first and then message, otherwise mom can't handle it.

There's a good reason for doing this, because some things have changed in the
meantime:
- Binaries are now always built in <CMAKE_BINARY_DIR>/bin.
- Configuration files are now always symlinked in <CMAKE_BINARY_DIR>/etc

The ConfigLocator searches the following default locations:
  ${LOFARROOT}, ${LOFARROOT}/etc, /opt/lofar, /opt/lofar/etc

If it doesn't find the configuration file in any of these directories, it will
also search the following directories:
  <BIN_DIR>, <BIN_DIR>/etc, <BIN_DIR>/.., <BIN_DIR>/../etc,
where <BIN_DIR> is the directory containing the executable.

And here's the problem. In the old days, the binary would be somewhere in the
build tree, in a location similar to its source. But now it's in
<CMAKE_BINARY_DIR>/bin. The extended search rule above will *always* find the
configuration file in <CMAKE_BINARY_DIR>/etc; there's no way you can override
this anymore. (Well, you could use the very dirty hack to redefine LOFARROOT
in your test, but that's plain ugly.)

So, by reintroducing the current directory in BASE_SEARCH_DIR, we now have
again the option to use a different configuration file for testing.

Task #10339: pipelines are also blocked if they are preceeded by approved/opened/error/obsolete status

Task #10339: apparently people change the status of (pre)scheduled/queued tasks in mom back to openend, so we should check these too and remove them if status=^Cenend in mom.

Task #10398: Adding INFO logging in the set-functions (mostly for setSplitter but also other set commands are not logged right now)

Task #10339: use factory method to get either a fast direct radb connection per thread, or get the default radb RPC connection. This improves loading times dramatically.

Task #10339: fixed type in getTasksTimeWindow. made check for tasks with conflicting resource_claims more efficient