CHG: Replace fatal MPI_Abort with Warning in GetHistoryFieldValue

[ayush4874]

Problem:
Currently, GetHistoryFieldValue in COutput.hpp triggers a fatal SU2_MPI::Error (causing an MPI_Abort) if a requested history field is missing. This behavior is overly aggressive for high-level interfaces (like the Python wrapper) or coupled simulations where a field might be temporarily unavailable or optional.

Solution:
Replaced the fatal error with a warning message (std::cout) and a default return value of 0.0. This ensures the simulation continues running while alerting the user.

Changes:

• Modified COutput.hpp: Replaced SU2_MPI::Error with a warning print and safe return in GetHistoryFieldValue.

[pcarruscag]

Can you share a config where you've run into this problem?

[ayush4874]

Hi @pcarruscag,

You can use any standard configuration file (like the quickstart_inviscid.cfg from the tutorial cases) to reproduce this.

The crash is not triggered by the configuration settings themselves, but by the API call when using the Python wrapper (or C++ driver) to request a field that doesn't exist.

Steps to Reproduce:

1. Run any standard case using the Python wrapper.
2. Request a non-existent history field inside the iteration loop.

import pysu2
# Use any valid config, e.g., quickstart_inviscid.cfg
driver = pysu2.CSinglezoneDriver("quickstart_inviscid.cfg", 1, MPI_COMM_WORLD)

# ... initialization ...

# CRASH: This triggers MPI_Abort immediately because "NonExistentVar" is not in the map.
# The user has no way to check for existence or catch the error.
val = driver.GetHistoryOutput("NonExistentVar")

Context:
In coupled simulations (like the ML-coupling I am working on), we need to query fields dynamically. If a field is missing (e.g., waiting for initialization), the current behavior (hard abort) kills the entire coupled simulation. A warning allows us to handle it safely.