…er + release v0.0.146

Closes #655 (Shape A was closed in v0.0.145; Shape B is this fix).

## Root cause

`vera/wasm/inference.py::_alias_array_element` followed alias
chains by checking `isinstance(target, ast.NamedType)` only.
When the alias target was a `RefinementType` (e.g.
`type NonEmptyArray = { @array<Int> | array_length(@array<Int>.0) > 0 }`),
the helper returned `None`.  Downstream `_infer_index_element_type`
then returned `None` for `@NonEmptyArray.0[0]`, the enclosing
`head` function got dropped via `[E602]` ("body contains
unsupported expressions — skipped"), and any call site to `head`
referenced a non-existent `$head` → `unknown func: $head` at WASM
validation.

## Fix

In `_alias_array_element`, peel any `RefinementType` layers from
the alias target before checking whether the base is a
`NamedType` pointing at `Array<T>`:

```python
if type_name in self._type_aliases:
    target = self._type_aliases[type_name]
    while isinstance(target, ast.RefinementType):
        target = target.base_type
    if isinstance(target, ast.NamedType):
        return self._alias_array_element(target.name, target.type_args)
```

Mirrors the existing `RefinementType` unwrap pattern in
`_canonical_named_type` (lines 619-622).  Refinement-of-Array
aliases now resolve their element type the same as a bare
`Array<T>`.

## Allowlist + KNOWN_ISSUES cleanup

`scripts/check_e602_clean.py` allowlist shrinks 6 → 5 entries
(the `head` entry is removed).  `KNOWN_ISSUES.md` `#655` row
removed entirely.

## Example update

`examples/refinement_types.vera::test_refine` previously called
only `safe_divide` and `to_percentage`.  Now also calls
`head([42, 1, 2])` so the `head` codepath is exercised
end-to-end by `scripts/check_examples.py` + `vera run`.  Returns
145 instead of 103 (`10/3 + 100` + `42`).

## Regression tests

New `TestHeadOverRefinement655ShapeB` class in
`tests/test_codegen.py` with two tests:

1. `test_head_over_refinement_compiles_and_runs` — asserts
   `(func $head ...)` appears in WAT (function not dropped via
   [E602]) AND that `execute()` returns 1 (`head([1, 2, 3])`).
2. `test_head_emits_no_e602_for_refinement_alias` — asserts no
   `[E602]` warning for `head` in the diagnostic stream.

Both fail against the pre-fix code path; both pass post-fix.

## Release prep (v0.0.146)

Bundled per the ABSOLUTE RULE (no separate cleanup PRs):

- `vera/__init__.py` + `pyproject.toml` + `uv.lock`: 0.0.145 → 0.0.146
- `CHANGELOG.md`: dated `[0.0.146] - 2026-05-12` heading + link refs
- `HISTORY.md` Stage 12 row: v0.0.146 line clarifying Shape A/B split
  on v0.0.145 + adding v0.0.146 row
- `docs/index.html`: version tag
- `README.md` / `TESTING.md` / `ROADMAP.md`: count refreshes
  (3,813 tests, 146 releases)
- `KNOWN_ISSUES.md`: removed #655 row (now fully closed); limitation
  count 34 (was 35)

## Verification

- mypy clean (59 source files)
- pytest: 3,799 passed, 14 skipped (was 3,797 + 2 new regression tests)
- All 86 conformance programs pass
- All 34 examples pass `vera check` + `vera verify`
- e602 gate: 116 files clean, 5 allowlist matched, 0 stale
- version-sync, doc-counts, limitations-sync: consistent

Refs #655 #659

Co-Authored-By: Claude <noreply@anthropic.invalid>

CR-1 (`HISTORY.md:334`) — release counter
The "Total: 810+ commits, 144 tagged releases, 58 active
development days." roll-up was 2 tags out of date.  Updated to
146 to reflect v0.0.145 + v0.0.146.  CodeRabbit also asked to
check ROADMAP.md for a matching counter — verified, ROADMAP
does not carry a tagged-release count (only test/example/chapter
counts).  Skipped that half.  `README.md:184` already had the
correct "146 releases" from the earlier release-prep update.

CR-2 (`scripts/check_e602_clean.py:98`) — allowlist note
Replaced `PR #<TBD>` with `PR #662` and rewrote the surrounding
prose to reflect that `head` is no longer an active allowlist
case.  The new note explains why it was removable (the
`_alias_array_element` `RefinementType` unwrap means
refinement-of-Array aliases resolve their element type the same
as bare `Array<T>`) and explicitly states `#655` is fully
closed.  Narrative + status now consistent.

Validation: e602 gate clean (5/5 matched, 0 stale); mypy clean.

Refs #655 #662

Co-Authored-By: Claude <noreply@anthropic.invalid>