fix: ExportMixin captures only outermost __init__ args for correct export/rebuild round-trips

[Copilot]

ExportMixin wrapped __init__ at every level of the class hierarchy and merged all args into one dict, so subclasses that derive a parent's constructor argument would export both their own arg and the parent's computed arg — breaking reconstruction.

class DerivedFieldNames(DataWriter):
    def __init__(self, value_field: str, **kwargs):
        super().__init__(field_names=["time_stamp", value_field], **kwargs)

writer = DerivedFieldNames(name="writer", value_field="capacity_mwh")
writer.export()["args"]
# Before: {'name': 'writer', 'value_field': 'capacity_mwh', 'field_names': [...]}
# → DerivedFieldNames(**args) raises TypeError: got multiple values for 'field_names'

# After: {'name': 'writer', 'value_field': 'capacity_mwh'}
# → rebuilds correctly

Changes

• plugboard/utils/export_mixin.py — _save_args_wrapper now sets a re-entrancy flag on the instance before calling __init__. Nested super().__init__() wrappers detect the flag and skip saving. The flag is deleted in a finally block so no stale attributes persist on instances.

• Fixes bug: export() captures __init__ arguments from every class in the chain, so subclasses cannot be rebuilt from a dumped config #272

[toby-coleman]

@copilot Can you add a minimal test for this

[github-actions]

Benchmark comparison for f452c7de (base) vs f38d1f23 (PR)

------------------------------------------------------------------------------------------------------------------ benchmark: 2 tests -----------------------------------------------------------------------------------------------------------------
Name (time in ms)                                                                         Min                 Max                Mean            StdDev              Median               IQR            Outliers     OPS            Rounds  Iterations
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
test_benchmark_process_run (main/.benchmarks/Linux-CPython-3.14-64bit/0001_base)     367.9127 (1.0)      380.3617 (1.01)     373.3145 (1.00)     5.0316 (1.76)     374.4442 (1.00)     7.5236 (2.66)          2;0  2.6787 (1.00)          5           1
test_benchmark_process_run (pr/.benchmarks/Linux-CPython-3.14-64bit/0001_pr)         368.6108 (1.00)     376.6151 (1.0)      372.7368 (1.0)      2.8621 (1.0)      373.0445 (1.0)      2.8255 (1.0)           2;0  2.6829 (1.0)           5           1
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Legend:
  Outliers: 1 Standard Deviation from Mean; 1.5 IQR (InterQuartile Range) from 1st Quartile and 3rd Quartile.
  OPS: Operations Per Second, computed as 1 / Mean

[Copilot]

@copilot Can you add a minimal test for this

Added two unit tests in tests/unit/test_component.py (commit test: add unit tests for ExportMixin outermost-args fix):

• test_export_captures_only_outermost_init_args — verifies that export()['args'] contains only the outermost constructor's arguments (not derived parent args), and that rebuilding from those args works.
• test_export_no_stale_flag_on_instance — verifies the re-entrancy flag is cleaned up and leaves no stale attributes on instances.

[github-actions]

Benchmark comparison for f452c7de (base) vs ac72ad5f (PR)

------------------------------------------------------------------------------------------------------------------ benchmark: 2 tests -----------------------------------------------------------------------------------------------------------------
Name (time in ms)                                                                         Min                 Max                Mean            StdDev              Median               IQR            Outliers     OPS            Rounds  Iterations
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
test_benchmark_process_run (pr/.benchmarks/Linux-CPython-3.14-64bit/0001_pr)         366.7561 (1.0)      379.0641 (1.0)      373.9058 (1.0)      4.6390 (1.0)      375.2914 (1.0)      5.7289 (1.0)           2;0  2.6745 (1.0)           5           1
test_benchmark_process_run (main/.benchmarks/Linux-CPython-3.14-64bit/0001_base)     382.7287 (1.04)     399.3943 (1.05)     388.1332 (1.04)     6.6391 (1.43)     385.4353 (1.03)     7.1702 (1.25)          1;0  2.5764 (0.96)          5           1
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Legend:
  Outliers: 1 Standard Deviation from Mean; 1.5 IQR (InterQuartile Range) from 1st Quartile and 3rd Quartile.
  OPS: Operations Per Second, computed as 1 / Mean