this should only be set on the target if using QuadMath.
every time cmake was rerun a build directory new entry was added,
leading to very long compile commands eventually.
The primary use case is identifying UDQ sets when the model defines
UDQs at the segment level (SU*). The call site will construct a
SegmentMatcher object backed by a specific ScheduleState, and then
segment-level UDQs will match against this to determine the correct
UDQ set on which to define the quantity.
Instead of listing a range where the errors are we now list the
location of each corresponding WELSEGS keyword to give the user all
the information that he may need.
Error looks like:
```
unknown location(0): fatal error: in "MissingCOMPSEGS": Opm::OpmInputError: Problem with keyword WELSEGS
In <memory string> line 38
Missing COMPSEGS keyword for the following multisegment well:
PROD01 in <memory string> at line 38.
```
This might not be 100% perfect but it will detect completely missing
COMPSEGS keywords for multisegment wells already during file parsing
instead of during the simulation at the report step where the problem
occurs (could be rather late). Hence it improves OPM/opm-simulators#4329.
We keep track of all the wells that appear in WELSEGS and COMPSEGS and
at the end of each report step parsed, we check that there is entry using
COMPSEGS for each well that was in WELSEGS.
Note that the keyword used for the error is the last one in the report
step, which is a bit misleading. Error message looks like:
```
Error: Problem with keyword WCONPROD
In /tmp/MSW-3D-03.DATA line 2462
Missing COMPSEGS keyword for well PROD01 between <memory string> line 0 and /tmp/MSW-3D-03.DATA line 2462.
Error: Unrecoverable errors while loading input: Problem with keyword WCONPROD
In /tmp/MSW-3D-03.DATA line 2462
Missing COMPSEGS keyword for well PROD01 between <memory string> line 0 and /tmp/MSW-3D-03.DATA line 2462.
```
If DISGASW is in the input deck gas in allowed to dissolved into water
Currently only works in combination with CO2STORE
For general support assosiated input tables needs to be provided
this way we don't have to drag 6MB of tables into multiple compile
units. some flexibility is lost (ie the template parameter for the
table), but this is unused in the current code.
comment out some templates that do not instance as they are referring
to non-existent table members.
it should be up to instance sites to make every available for
the templates to instance properly, even though we know we are
instancing over Evaluations.
