fix(postgresql): stop declaring IPNetwork on both the cidr and inet mappings (weasel#405)

[jeremydmiller]

Closes #405.

PostgresqlProvider.GetTypeMapping breaks a CLR-type tie with LastOrDefault:

NpgsqlTypeMapper.Mappings.LastOrDefault(mapping => mapping.ClrTypes.Contains(type));

LastOrDefault only means something over an ordered sequence. Mappings is a JasperFx.Core.Cache backed by an ImHashMap, so it enumerates in hash-tree order:

enumeration order : <custom>, <custom>, Bigint, Boolean, Box, Bytea, Circle, Char
source declares   : Smallint, Integer, Bigint, Real, Double, Numeric, Money, Text

IPNetwork was the one CLR type claimed by two mappings — Cidr (line 85) and Inet (line 86):

mappings claiming IPNetwork : Inet/inet  |  Cidr/cidr   (LastOrDefault wins)
GetDatabaseType(IPNetwork)  : cidr

cidr is the right answer and ipnetwork_resolves_to_cidr passes — but only by accident. The source declares Cidr before Inet, so under genuine insertion order LastOrDefault would return inet and that test would fail today. It passes because the hash layout happens to enumerate Cidr last, which is a function of the current key set — and consuming code is explicitly invited to add mappings.

Change

Drop the stray typeof(IPNetwork) from the Inet mapping. IPAddress, NpgsqlInet and the (IPAddress, int) tuple stay on Inet. Resolution becomes unambiguous and order-independent.

Two guards so this cannot silently come back:

• ipnetwork_is_claimed_by_exactly_one_mapping
• no_clr_type_is_claimed_by_more_than_one_mapping — general, reports offenders as Type <- Mapping, Mapping

Counterfactual — restoring the duplicate fails both:

PostgresqlProviderTests.ipnetwork_is_claimed_by_exactly_one_mapping [FAIL]
PostgresqlProviderTests.no_clr_type_is_claimed_by_more_than_one_mapping [FAIL]

Verification

All four CI matrix legs (net9.0/net10.0 × case-sensitive true/false): 788 passed, 0 failed, 3 skipped.

One unrelated intermittent failure surfaced during validation — DatabaseWithTablesTests.detect_and_apply_schema_changes, which races other collections over the shared public schema. Confirmed pre-existing on clean master (reproduced in 3 runs there) and filed separately as #407. It is not affected by this change.

🤖 Generated with Claude Code