Bug Description: Results for at least one example depend on the geometry input (directly in .yaml or via exodus read)

During my update of the solid tutorial (see PR #694) I first noticed that reading the HEX27 elements needs special treatment, which in the meantime has been implemented by @sebproell in PR #712 (thanks for that!).

Now it turns out that the results are different depending on the kind of geometry input.

Expectations

Of course the results should be the same no matter whether the exodus input is read by pre_exodus or 4C itself. At least on my computer and on our cluster it's not.

Steps to Reproduce

You may verify this behavior by the following procedure:

1. generate the 4C input file of the solid tutorial, e.g., by running ctest -FC cleanup -R tutorial_solid
   (File framework_test_output/tutorial_solid/xxx.4C.yaml is being created), the test should run successfully.
2. Replace all lines from DNODE-NODE TOPOLOGY: [] till the end by:

STRUCTURE GEOMETRY:
  FILE: "tutorial_solid.e"
  ELEMENT_BLOCKS:
    - ID: 1
      ELEMENT_NAME: SOLID
      ELEMENT_DATA: "MAT 1 KINEM nonlinear"

3. Copy the exodus file tutorial_solid.e from the tests/framework_test/tutorial_solid.e to the directory, where the test ran, or replace the FILE option inside STRUCTURE GEOMETRY with the full path name.
4. Add the option ENTITY_TYPE: node_set_id to each boundary condition entry (4 times in DESIGN SURF DIRICH CONDITIONS). Otherwise an error will be thrown immediately (actually I don't know why this is necessary depending on the geometry read option).
5. Save the yaml input file.
6. Run the problem again (not the ctest!).

An error is thrown, because the results are different:

STRUCTURE: structure   dispy   (elongation) at node 2477         is WRONG --> actresult= 4.56117520898904338e-01, givenresult= 4.56123591074363621e-01, abs(diff)= 6.07017545928290048e-06 > 9.99999999999999980e-13
STRUCTURE: structure   dispx   (necking) at node   6     is WRONG --> actresult=-4.11160719183618420e-02, givenresult=-4.25664885567924312e-02, abs(diff)= 1.45041663843058916e-03 > 9.99999999999999980e-13

Your Environment

• **Relevant repo SHA1s:d476353
• **Relevant configure flags or configure script:RELEASE
• **Operating system and version:Ubuntu 24.04 LTS
• **Compiler and TPL versions:gcc 13.3

Same behavior comes up on the cluster (same repo SHA1, CentOS 7.9, gcc 13.2)

Related Issues and Pull Requests

Blocks #694