test(parity): expected-exit=1 for 3 intentional-throw expected-output tests

[proggeramlug]

Problem

run_parity_tests.sh compares an expected-output test's process exit code
against test-parity/expected-exit/<name>.txt, defaulting to 0 when that file
is absent. Three Perry-specific expected-output tests intentionally end by
throwing a TypeError (verifying spec behavior) — their committed
test-parity/expected/<name>.txt snapshots already capture the thrown error —
but had no expected-exit entry. So the harness expected exit 0 while
Perry (correctly, matching node --experimental-strip-types) exits 1,
producing a false parity failure even though Perry's stdout+error and exit
already match Node.

Affected: test_issue_462_nullish_property_access,
test_issue_510_primitive_method_typeerror,
test_decorators_replacement_unsupported.

Fix

Add test-parity/expected-exit/<name>.txt = 1 for all three.

No runtime/compiler change — Perry's behavior was already correct; this only
corrects the harness's expected exit code for these throw-on-purpose cases.

Verification

Each now passes:

./run_parity_tests.sh --filter test_issue_462_nullish_property_access  → PASS (expected-output)
./run_parity_tests.sh --filter test_issue_510_primitive_method_typeerror → PASS (expected-output)
./run_parity_tests.sh --filter test_decorators_replacement_unsupported → PASS (expected-output)