[CONTRACTS] DFCC loop assigns infererence with functions inlined

[qinheping]

Current loop assigns inference of DFCC can only infer targets considering only instructions in the same function. However, variables may be written to in the call of another function from function call inside the loop. So in this PR, we infer loop assigns based on a copy of the function and inlining all calls in the copy.

• Each commit message has a non-empty body, explaining why the change was made.
• Methods or procedures I have added are documented, following the guidelines provided in CODING_STANDARD.md.
• The feature or user visible behaviour I have added or modified has been documented in the User Guide in doc/cprover-manual/
• Regression or unit tests are included, or existing tests cover the modified code (in this case I have detailed which ones those are in the commit message).
• My commit message includes data points confirming performance improvements (if claimed).
• My PR is restricted to a single feature or bugfix.
• White-space or formatting changes outside the feature-related changed lines are in commits of their own.

[codecov]

Codecov Report

Attention: Patch coverage is 84.69388% with 15 lines in your changes missing coverage. Please review.

Project coverage is 78.80%. Comparing base (018c61c) to head (0444fad).

Files with missing lines	Patch %	Lines
...ntracts/dynamic-frames/dfcc_infer_loop_assigns.cpp	84.70%	13 Missing ⚠️
...trument/contracts/dynamic-frames/dfcc_cfg_info.cpp	84.61%	2 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #8490      +/-   ##
===========================================
- Coverage    78.90%   78.80%   -0.11%     
===========================================
  Files         1727     1727              
  Lines       198425   198785     +360     
  Branches     18523    18529       +6     
===========================================
+ Hits        156561   156643      +82     
- Misses       41864    42142     +278     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[qinheping]

We also switch to a more aggressive loop-assigns widening strategy that also widens all dereference to whole object. The reason to do so is that it fits better for Kani loop contracts, while we have a good loop assigns specification support for C loop contracts for which the inference is less needed.

[tautschnig]

Is this change now strictly necessary given the additional widening?

[qinheping]

Yes, new addition widening will infer object_whole(data), i which is not accurate for this test.