Enhancement on logging message

[jinningwang]

Is your feature request related to a problem? Please describe.
In file loading stage, some logging messages are repetitive, especially when the case size is large.

Describe the solution you'd like

• In file loading stage, log only one time for the parameter correction for devices in one same model.
• In TDS initialization process, skip messaging “out of typical range” for offline devices

Describe alternatives you've considered
A clear and concise description of any alternative solutions or features you've considered.

Additional context
One message for each, but actually we can mention just one time for a Model.

Non-zero parameter GENROU.M corrected to 6
Non-zero parameter GENROU.M corrected to 6
Non-zero parameter GENROU.M corrected to 6
Non-zero parameter GENROU.M corrected to 6
Non-zero parameter GENROU.M corrected to 6
Non-zero parameter GENROU.M corrected to 6
Non-zero parameter GENROU.M corrected to 6
Non-zero parameter GENROU.M corrected to 6
...

[cuihantao]

This has been addressed on the v2.0-dev branch.

How it works now: Instead of logging one warning per device immediately during data loading, NumParam.add() now returns a violation type string (e.g., non_zero) without logging anything. ModelData.add() collects these into a _param_corrections dict keyed by (param_name, violation_type) → list of device indices. Then during System.setup(), after find_devices() completes (so auto-added devices are included), _report_param_corrections() iterates all models and calls report_corrections(), which emits one grouped warning per unique violation:

GENROU: 10 device(s) had non_zero param <M> corrected to 6. idx=['GEN1', 'GEN2', ..., 'GEN10']

The dict is cleared after reporting.

Regarding the second item (skip "out of typical range" for offline devices) — I didn't find specific handling for that in the current code, so that may still be open.

Fixed in commit 925297c4 ("Batch parameter correction reporting with deferred setup-time warnings").