The getActiveElements function of a process is used from
now on (as it is correctly handling multiple variable deactivations)
and with this the process_id and process_variable are no longer
needed as arguments.

Slightly change the test to see influence of deactivated temperature
field on the pressure variable.

This fixes the issue that the assembly is not executed when
the first process' variable is deactivated.

Small fixes

See merge request ogs/ogs!5020

Improved error handling, mostly for Python BCs/STs

See merge request ogs/ogs!5019

The Simulation instance needs to be alive upon Python exception
handling, otherwise e.what() will lead to a segfault.
Why? → Because Simulation "owns" the Python interpreter.

... this should improve the compilation error reporting

The original input is not close enough to unity.

[web] Updated gcc install instructions

See merge request ogs/ogs!5018

See https://discourse.gohugo.io/t/how-to-comment-out-shortcodes-in-markdown/14893/10

Follow-up on !5011.

Further RichardsMechanics refactoring

See merge request ogs/ogs!5012

Provide MKL container

See merge request ogs/ogs!5011

Christoph Lehmann authored 9 months ago and Dmitry Yu. Naumov committed 9 months ago

... from the local assembler to the interface classe, like in TRM.

Enable reflected I/O for StrongTypes

See merge request ogs/ogs!5004

Enable Volumetric and Python source terms on 0D elements (aka Points)

See merge request ogs/ogs!5009

Computing shape matrices of a d-dimensional Element in less than d
dimensions is not meaningful.

With that change "Volumetric"/"Python" source terms can be applied to nodes, as well.
Thereby, the special "Nodal" source term has become obsolete, now.