fix one-in forecaster map wiring and remove dead forecast map code

[xmariachi]

Purpose of Changes and their Description

Summary

• Use singleForecastImpliedInference in calcOneInValue so one-in computation is scoped to the selected forecaster.
• Remove unused one-in plumbing (singleForecasterToForecast and unused internal arg field).
• Add a regression test for nil forecast-implied entries in one-in forecaster inference path.

Validation
go test ./x/emissions/keeper/inference_synthesis -count=1

Link(s) to Ticket(s) or Issue(s) resolved by this PR

Are these changes tested and documented?

• If tested, please describe how. If not, why tests are not needed.
  Tests pass, added test
• If documented, please describe where. If not, describe why docs are not needed.
  no need, no func change
• Added to Unreleased section of CHANGELOG.md?
  no need, no func change

---

Summary by cubic

Aligns one-in forecaster computation to the selected forecaster’s forecast-implied inference and remove dead forecast map wiring. Addresses Linear ENGN-5777 by fixing map wiring and hardening nil-entry handling.

• Bug Fixes

  • Pass singleForecastImpliedInference into calcWeightedInference for one-in.
  • Guard against nil entries when building the single-forecaster implied inference map.
  • Add regression test to confirm nil forecast-implied entries do not panic.

• Refactors

  • Remove singleForecasterToForecast and the ForecasterToForecast arg from calcOneInValueArgs.
  • Simplify one-in path inputs; existing epoch outputs remain unchanged.

^{Written for commit e34054d. Summary will update on new commits.}

[greptile-apps]

Your free trial has ended. If you'd like to continue receiving code reviews, you can add a payment method here.

[cubic-dev-ai]

cubic analysis

No issues found across 2 files

Linked issue analysis

Linked issue: ENGN-5777: inference_synthesis: clean one-in forecaster map wiring and remove dead local forecast map

Status	Acceptance criteria	Notes
⚠️	One-in forecaster outputs remain unchanged for existing epoch fixture tests.	Existing epoch test remains but no explicit regression check shown
✅	No panic when a forecaster key in ForecasterToForecastImpliedInference has a nil value.	Added test that sets a nil entry and asserts NotPanics
✅	singleForecasterToForecast no longer exists in calcOneInValue.	Removed creation and use of singleForecasterToForecast
⚠️	Internal one-in call path compiles and tests pass.	Call wiring updated but test run / CI pass not shown in diffs
✅	Pass singleForecastImpliedInference into calcWeightedInference for one-in computation.	One-in path now uses singleForecastImpliedInference in args
✅	Remove unused local singleForecasterToForecast.	Local singleForecasterToForecast map creation removed
✅	Remove unused internal arg field ForecasterToForecast from calcOneInValueArgs and call wiring.	Field removed from struct and removed where args are assembled
✅	Add nil-safe construction for singleForecastImpliedInference to avoid present-but-nil map entry hazards.	Checks ok && inferred != nil before adding to map
✅	Add regression test for one-in path nil-entry handling.	New test verifies nil-entry handling and ensures no panic
⚠️	Preserve behavior for existing valid inputs while hardening edge handling.	Code intends to preserve behavior but diffs lack explicit regression output comparison

Architecture diagram

sequenceDiagram
    participant Manager as Inference Synthesis Manager
    participant Loop as GetOneInForecasterInferences()
    participant Calc as calcOneInValue()
    participant Weights as calcWeightedInference()

    Note over Manager,Weights: One-In Forecaster Evaluation Flow

    Manager->>Loop: Get inferences for all forecasters
    
    loop for each OneInForecaster in Forecasters
        Loop->>Calc: Calculate value for specific forecaster
        
        Calc->>Calc: NEW: Initialize singleForecastImpliedInference map
        
        alt NEW: Forecaster has valid implied inference
            Calc->>Calc: Add forecaster's inference to isolated map
        else Forecaster entry is nil or missing
            Note right of Calc: Regression fix: Prevent panic by skipping map insertion
        end

        Calc->>Calc: Isolate singleForecasterRegret and worker list
        
        Calc->>Weights: CHANGED: Call with scoped singleForecastImpliedInference
        Note over Calc,Weights: Previously passed full global map; now restricted to current forecaster
        
        Weights-->>Calc: Weighted result for this forecaster
        Calc-->>Loop: One-in value
    end

    Loop-->>Manager: Return collection of one-in inferences

Loading

[zale144]

I believe I already fixed this in the classification PR. let's leave this for later, we could still use the unit test