transform_pmf_model_out() broken by new scoringutils

[zkamvar]

I just created a repo that will run weekly integration tests and found an error in hubEvals where four tests are failing:

 ══ Failed tests ════════════════════════════════════════════════════════════════
  ── Error ('test-score_model_out.R:329:3'): score_model_out succeeds with valid inputs: nominal pmf output_type, default metrics, custom by ──
  Error in `scoringutils::as_forecast_nominal(data, forecast_unit = c("model", 
      task_id_cols), observed = "observation", predicted = "value", 
      model = "model", predicted_label = "output_type_id")`: unused argument (model = "model")
  Backtrace:
      ▆
   1. └─hubEvals::score_model_out(...) at test-score_model_out.R:329:3
   2.   └─hubEvals:::transform_pmf_model_out(...) at hubEvals/R/score_model_out.R:96:3
  ── Error ('test-transform_pmf_model_out.R:16:3'): transform_pmf_model_out succeeds with valid inputs ──
  Error in `scoringutils::as_forecast_nominal(data, forecast_unit = c("model", 
      task_id_cols), observed = "observation", predicted = "value", 
      model = "model", predicted_label = "output_type_id")`: unused argument (model = "model")
  Backtrace:
      ▆
   1. └─hubEvals:::transform_pmf_model_out(...) at test-transform_pmf_model_out.R:16:3
  ── Error ('test-transform_pmf_model_out.R:46:3'): transform_pmf_model_out doesn't depend on specific column names for task id variables ──
  Error in `scoringutils::as_forecast_nominal(data, forecast_unit = c("model", 
      task_id_cols), observed = "observation", predicted = "value", 
      model = "model", predicted_label = "output_type_id")`: unused argument (model = "model")
  Backtrace:
      ▆
   1. └─hubEvals:::transform_pmf_model_out(...) at test-transform_pmf_model_out.R:46:3
  ── Error ('test-transform_pmf_model_out.R:59:3'): transform_pmf_model_out throws an error if model_out_tbl has no rows ──
  Error in `scoringutils::as_forecast_nominal(data, forecast_unit = c("model", 
      task_id_cols), observed = "observation", predicted = "value", 
      model = "model", predicted_label = "output_type_id")`: unused argument (model = "model")
  Backtrace:
      ▆
   1. ├─testthat::expect_error(...) at test-transform_pmf_model_out.R:59:3
   2. │ └─testthat:::expect_condition_matching(...)
   3. │   └─testthat:::quasi_capture(...)
   4. │     ├─testthat (local) .capture(...)
   5. │     │ └─base::withCallingHandlers(...)
   6. │     └─rlang::eval_bare(quo_get_expr(.quo), quo_get_env(.quo))
   7. ├─base::suppressWarnings(...)
   8. │ └─base::withCallingHandlers(...)
   9. └─hubEvals:::transform_pmf_model_out(...)
  
  [ FAIL 4 | WARN 0 | SKIP 0 | PASS 64 ]

Source

The source of this is epiforecasts/scoringutils#915, where the model parameter was removed from as_forecast_nominal()