fix(coverage): make coverage analyze fail when tests fail

The coverage analyze command now properly propagates test failures while still
generating the coverage report. Previously, it would ignore test failures and
return success if the coverage report was generated successfully.

Behavior:
- Always generate the coverage report (useful for debugging)
- Propagate the test result exit code (2 for test failures)
- Only return an error if the coverage report itself fails

Includes a regression test with a failing test case to prevent future
regressions.

Author: peter-jerry-ye

crates/moon/src/cli/coverage.rs

@@ -25,7 +25,7 @@ use clap::Parser;
 use moonutil::dirs::PackageDirs;
 use walkdir::WalkDir;
 
-use super::{TestSubcommand, UniversalFlags, run_test};
+use super::{run_test, TestSubcommand, UniversalFlags};
 
 #[derive(Debug, clap::Parser, Default)]
 #[clap(
@@ -102,17 +102,15 @@ fn run_coverage_analyze(
         test_flags,
     )?;
 
-    if test_result != 0 {
-        return Ok(test_result);
-    }
-
     let mut report_flags = CoverageReportSubcommand::default();
     report_flags.args.push("-f=simp_caret".into());
     if let Some(package) = &args.package {
         report_flags.args.push(format!("-p={package}"));
     }
     report_flags.args.extend(args.extra_flags);
-    run_coverage_report(cli, report_flags)
+    let _ = run_coverage_report(cli, report_flags)?;
+
+    Ok(test_result)
 }
 
 fn run_coverage_clean(cli: UniversalFlags) -> Result<i32, anyhow::Error> {

crates/moon/tests/test_cases/moon_coverage.rs

@@ -137,4 +137,5 @@ fn test_moon_coverage_analyze_fails_on_test_failure() {
         .failure();
     let stdout = String::from_utf8(output.get_output().stdout.to_owned()).unwrap();
     assert!(stdout.contains("Total tests") && stdout.contains("failed"));
+    assert!(stdout.contains("All source files are fully covered"));
 }