Random testing and tweak optimisation on by default

[springfall2008]

No description provided.

[Copilot]

Pull request overview

Adds a reproducible “random scenario” benchmarking workflow to exercise Predbat’s optimiser from the coverage/ directory, and changes the optimisation flow so that when the full second pass is disabled Predbat runs a faster tweak pass by default (removing the old calculate_tweak_plan switch).

Changes:

• Introduces random scenario generation/run/compare/profile commands to apps/predbat/unit_test.py, plus supporting docs and helper scripts under coverage/.
• Removes the calculate_tweak_plan configuration option and wires tweak_plan() as the default “fast” fallback when calculate_second_pass is off.
• Includes small optimiser/prediction adjustments and updated expected outputs/baseline artifacts for the new workflow.

Reviewed changes

Copilot reviewed 13 out of 18 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
mkdocs.yml	Adds the new random scenarios documentation page to the MkDocs nav.
docs/random-scenarios.md	Documents the random scenario benchmarking workflow and CLI options.
docs/customisation.md	Removes the old calculate_tweak_plan switch documentation.
coverage/run_random_profile	Convenience script for profiling a scenario run.
coverage/run_random	Convenience script for running and comparing random scenario results.
coverage/cases/random_results.json	Baseline random-run results JSON for comparisons.
coverage/cases/predbat_debug_pre_saving1.yaml.expected.json	Updates expected debug-case output after optimisation behavior changes.
apps/predbat/unit_test.py	Adds CLI flags for random scenario generate/run/compare/profile and skips random_scenarios.yaml in debug cases.
apps/predbat/tests/test_single_debug.py	Removes now-obsolete commented calculate_tweak_plan toggles.
apps/predbat/tests/test_random_scenarios.py	Adds the random scenario framework implementation (generation, execution, comparison, profiling).
apps/predbat/tests/test_infra.py	Removes calculate_tweak_plan from the default test config.
apps/predbat/prediction.py	Removes a SOC rounding step within prediction.
apps/predbat/predbat.py	Removes calculate_tweak_plan reset initialization.
apps/predbat/plan.py	Adjusts optimiser heuristics, reduces copying overhead, changes tweak/second-pass flow, and makes tweak_plan return metrics.
apps/predbat/fetch.py	Stops reading the removed calculate_tweak_plan config option.
apps/predbat/config.py	Removes calculate_tweak_plan from CONFIG_ITEMS.
.cspell/custom-dictionary-workspace.txt	Adds spellings used by the new profiling/CLI docs.

Comments suppressed due to low confidence (1)

docs/customisation.md:256

• The documentation for switch.predbat_calculate_second_pass now implies that turning it off disables the “second pass” entirely, but the code falls back to running tweak_plan() when calculate_second_pass is false. Please update this section to explain the new behavior (full second pass vs faster tweak pass) so users understand what happens when the switch is off.

**switch.predbat_calculate_second_pass** (_expert mode_) When turned On causes Predbat to perform a second pass optimisation across all the charge and export windows in time order.

Note: This feature is quite slow so may need a higher-performance machine so is turned Off by default.

This can help to slightly improve the plan for tariffs like Agile but can make it worse in some fixed rate tariffs in which you want to force export late.