feat: resolve bats load as a sourcing command

[kolyshkin]

Fixes #181.

Problem

Bats test files pull in helper files with load test_helper, not source. Since load was not recognized as a sourcing command, none of the helper's symbols were reachable from a .bats file — go to definition on a helper function returned nothing.

The only workaround today is bashIde.includeAllWorkspaceSymbols, which is a rather blunt instrument: it widens completion and definition across the entire workspace, well beyond what the file actually pulls in.

Change

load is now treated as a sourcing command, but only in .bats files — load is a common enough name for an unrelated command or function elsewhere. Path resolution follows what bats itself does: relative to the directory of the test file, retrying with a .bash suffix when the given path is not a file.

#!/usr/bin/env bats

load test_helper        # -> ./test_helper.bash
load ./test_helper.bash # -> ./test_helper.bash
load /abs/path/helper   # -> /abs/path/helper.bash

bats_load_library is deliberately left out: it resolves against BATS_LIB_PATH, which is not statically known, so it would only produce unresolvable source commands (and, with enableSourceErrorDiagnostics, spurious diagnostics).

Tests

• sourcing.test.ts: resolution of the forms above, plus the dynamic/unfinished cases, and a test asserting load is not treated as sourcing outside .bats files.
• analyzer.test.ts: an end-to-end findDeclarationLocations test going from a .bats fixture into its helper — with includeAllWorkspaceSymbols off, which is exactly the case that did not work before.
• New fixtures under testing/fixtures/bats/. The added .bash fixture bumps FIXTURE_FILES_MATCHING_GLOB and adds one empty entry to two workspace-wide rename snapshots.

Note: formatter › should format when shfmt is present fails on my machine both before and after this change — the expected-message regex does not match shfmt 3.13.1's wording. Unrelated to this PR.

[kolyshkin]

@skovhus I think you need to enable CI here (there should be a button). If not possible, close and reopen this PR does the trick.

[kolyshkin]

rebased on top of merged #1393; CI started!

[kolyshkin]

CI is green now but auto-merge is now disabled because of rebase :( PTAL @skovhus

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 80.65%. Comparing base (d14b5d2) to head (61b5428).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1392      +/-   ##
==========================================
+ Coverage   80.56%   80.65%   +0.08%     
==========================================
  Files          29       29              
  Lines        1513     1520       +7     
  Branches      374      376       +2     
==========================================
+ Hits         1219     1226       +7     
  Misses        235      235              
  Partials       59       59              

☔ View full report in Codecov by Harness.
📢 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.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.