Stability issues when running a large number of mechanical solves in one instance

Note: This was reported for the embedding interface, but the underlying issue also affects both the remote interface and Mechanical runs without PyMechanical.

When running a large number of solves in one instance, eventually there will be a crash. For example:

for in range(1000):
model.Solve()
model.ClearGeneratedData()

This is likely due to memory leaks. It is recommended for workflows of this nature to split into multiple python processes.

No files, tests, or entry points are named. Start by reproducing the 1000-solve loop and measuring memory and crash behavior across the embedding, remote, and standalone Mechanical interfaces. Done means identifying and fixing the leak or crash cause, with evidence that long-running solve workflows remain stable.

Written by the indexing model from the issue text.