Fix #1043: Add print method for multivariate forecasts showing joint_across

[nikosbosse]

CLAUDE:

Summary

• Adds print.forecast_multivariate_sample() and print.forecast_multivariate_point() S3 methods that display which columns are jointly forecasted
• Both methods delegate to a shared print_multivariate_forecast() helper that computes joint_across as setdiff(get_forecast_unit(x), get_grouping(x)) and shows it between forecast unit and data table output
• Fixes the original method name from print.forecast_sample_multivariate (old class name) to print.forecast_multivariate_sample (current class name)

Root cause

No dedicated print methods existed for multivariate forecast objects — they fell through to the generic print.forecast(), which only shows forecast type and forecast unit but has no awareness of joint_across or the .mv_group_id grouping structure. Additionally, the original PR had the method named after the old class (forecast_sample_multivariate) so it was never dispatched.

What the fix does

Creates print_multivariate_forecast() in R/class-forecast-multivariate-sample.R that:

1. Prints forecast type and forecast unit (matching print.forecast() behavior)
2. Computes joint_across columns from the difference between forecast unit and grouping columns
3. Displays "Joint across:" followed by the column names
4. Prints the data table via print(as.data.table(x))
5. Returns the input invisibly

Both print.forecast_multivariate_sample() and print.forecast_multivariate_point() are thin wrappers that delegate to this helper.

Test plan

• Snapshot tests for both sample and point print output
• Returns object invisibly (both types)
• Output contains all expected sections (type, unit, joint_across)
• Error branches covered (mocked get_forecast_type and get_forecast_unit)
• Full test suite passes (766 tests, 0 failures)

Closes #1043

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.00%. Comparing base (d4bbd6e) to head (6430464).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1074      +/-   ##
==========================================
+ Coverage   97.98%   98.00%   +0.01%     
==========================================
  Files          37       37              
  Lines        1984     2001      +17     
==========================================
+ Hits         1944     1961      +17     
  Misses         40       40              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[nikosbosse]

CLAUDE: Approve (posted as comment due to GitHub self-review restriction). Clean, focused implementation of print.forecast_sample_multivariate() that correctly computes joint_across via setdiff(get_forecast_unit(x), get_grouping(x)) and displays it in the print output. All 5 tests pass and match specs. Three minor notes: (1) the method duplicates ~30 lines from print.forecast() rather than using NextMethod() — deliberate to insert the "Joint across:" line, but creates maintenance coupling; (2) @importFrom cli cli_inform is missing from the roxygen tag (works via package-level NAMESPACE but not best practice); (3) Test 5 name says "still calls parent print.forecast()" but the implementation replicates the parent's logic instead of calling it — the test itself is valid, just the name is misleading.

[nikosbosse]

CLAUDE: Here's an example of the new print output for example_multivariate_sample:

> print(example_multivariate_sample)
Forecast type: multivariate_sample
Forecast unit:
location, location_name, target_end_date, target_type, forecast_date, model,
and horizon
Joint across:
location and location_name

       location location_name target_end_date target_type forecast_date
         <char>        <char>          <Date>      <char>        <Date>
    1:       DE       Germany      2021-01-02       Cases          <NA>
    2:       DE       Germany      2021-01-02      Deaths          <NA>
    3:       DE       Germany      2021-01-09       Cases          <NA>
    ...
35624: epiforecasts-EpiNow2       2  24.43419        40       78          260

The new "Joint across" line shows which columns vary within each multivariate group (i.e., location and location_name are the dimensions being jointly forecasted), which is the key information that was previously invisible when falling through to the generic print.forecast().

Also pushed a fix for the lint errors (fixed = TRUE in grepl calls) and added tests for the error branches to address the codecov gap.

[nikosbosse]

@seabbs I think adding joint across to the print output would be nice. Things we could consider in the future

• the multivariat forecast class
• extracting the joint across code into a helper