[SG-4917] feat(infracost): name the cost being measured and what it covers

[refeed]

What

A cost message was the comparison and nothing else:

1. FAILED: `300.1` is not less than or equal to `20`
2. FAILED: `100.1` is not less than or equal to `-1`

It says neither which cost was measured nor what it covered — a monthly and an hourly figure of the same size read identically, and you have to go open the policy to tell which one you are looking at. Now:

1. FAILED: [all resources (2 resources)] total_monthly_cost: `300.1` is not less than or equal to `20`
2. FAILED: [aws_eks_cluster, aws_s3_bucket (1 resource)] total_monthly_cost: `100.1` is not less than or equal to `-1`

Same provider context mechanism terraform_plan uses, so the detail is also structured:

{
  "operation_type": "total_monthly_cost",
  "label": "aws_s3_bucket",
  "attribute": "total_monthly_cost",
  "resource_type": ["aws_s3_bucket"],
  "matched_resources": 1,
  "qualifier": "1 resource",
  "currency": "USD"
}

The resource count is there for a specific trap

resource_type is matched against the infracost breakdown, so a typo produces a cost of 0 — and LessThan 20 passes, while measuring nothing at all:

before:  1. PASSED: `0` is less than `20`
after:   1. PASSED: [aws_instances (0 resources)] total_monthly_cost: `0` is less than `20`

The verdict is deliberately unchanged — a policy that passed still passes, since changing that would turn existing green pipelines red. But the reason is now visible instead of a green check on an empty measurement. test_a_resource_type_matching_nothing_says_so_instead_of_reporting_a_bare_zero pins both halves of that.

Also in here

• format_context_prefix gained a qualifier key, rendering in the same parenthetical as action, for providers that have no notion of a planned action. terraform_plan keeps using action; nothing about its output changes.
• The core's non-ProviderError error branch (added in 1.1.0, core.py:118) was the one message path still not getting the context prefix. Now it does, so [all resources] total_monthly_cost: 'projects not found in input_data' names the evaluator it belongs to.
• _is_every_resource() collects the three spellings of "all resources" (None, "*", ["*"]) that provide already accepted, so the message agrees with what was actually measured on every path.

Verification

• pytest: 787 passed. The 11 failures are pre-existing and need a terraform binary on PATH.
• 9 new tests in tests/providers/infracost/test_infracost_message_context.py, including one asserting the two cost walks still return the same totals (they now return (total, matched_count)).
• black --check clean under both 25.9.0 and 26.5.1, and this branch adds zero files to the set main already flags under the newer one.
• pydocstyle clean.
• E2E on QA to follow — the terraform step template runs infracost, so a real run in demo-org exercises this.

Two things found while doing this, not fixed here

• tests/providers/infracost/tests_infracost.py is not collected and is vacuous. The filename is tests_ rather than test_, so pytest skips it. Its 6 tests would pass regardless: they pass costType instead of operation_type, so provide raises KeyError, returns {"value": None}, and every assertion is None != 0. Left alone — renaming it would surface 6 failures unrelated to this PR.
• The new test file is test_infracost_message_context.py rather than test_message_context.py because pytest cannot collect two same-named test modules without __init__.py in the test dirs.

Related: SG-4917, tirith#243

[notion-workspace]

Tirith: include resource context in evaluation result messages

[codecov]

Codecov Report

❌ Patch coverage is 97.72727% with 1 line in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
src/tirith/providers/infracost/handler.py	97.36%	0 Missing and 1 partial ⚠️

Files with missing lines	Coverage Δ
src/tirith/core/core.py	85.77% <100.00%> (+0.26%)	⬆️
src/tirith/providers/common.py	98.48% <100.00%> (ø)
src/tirith/providers/infracost/handler.py	86.25% <97.36%> (+16.43%)	⬆️

🚀 New features to boost your workflow:

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

[sonarqubecloud]

❌ The last analysis has failed.

See analysis details on SonarQube Cloud