feat(sqlserver): Part 7 — adapter_impl.rs match arms

[axellpadilla]

Stacked on #16 (Part 6).

Resolves all 33 E0004 non-exhaustive-match errors in adapter_impl.rs — the last remaining compile-error backlog blocking a full dbt-adapter crate build for SQL Server. Each site got a SqlServer arm or was folded into an existing group, verified against v1 dbt-sqlserver and the shared macro packages rather than assumed identical to Fabric.

Where it matches Fabric exactly

• valid_incremental_strategies: same four strategies — v1's SQLServerAdapter.valid_incremental_strategies() returns the identical list.
• list_schemas_inner's "schema" column name: confirmed via v1's sqlserver__list_schemas, which selects name as [schema].
• Every purely BigQuery- or Databricks-specific method (partition config, dataset location, table options, DBR capabilities, Iceberg helpers, describe_dynamic_table, etc.): SqlServer joins the existing "unimplemented" bucket alongside Fabric and the other non-BigQuery/non-Databricks adapters.

Where it diverges

• get_constraint_support: v1's SQLServerAdapter.CONSTRAINT_SUPPORT overrides all five constraint types to Enforced — including Check, unlike Fabric which marks Check NotSupported. Given its own block rather than folded into Fabric's.
• standardize_grants_dict: joins the Postgres/Bigquery/DuckDB/Alt group (grantee/privilege_type columns), not the "grants not implemented" bucket. v1's sqlserver__get_show_grant_sql selects exactly those two column names, and SQLServerAdapter doesn't override standardize_grants_dict, so it falls through to the same base implementation those adapters use.
• truncate_relation and get_columns_in_relation's macro-dispatch group: v1 has sqlserver__truncate_relation and sqlserver__get_columns_in_relation macros, so SqlServer joins the macro-based groups rather than any unimplemented bucket.
• verify_database: joins the "unimplemented" bucket (Fabric/Snowflake/Databricks/Salesforce) rather than the Postgres/DuckDB/Alt/ClickHouse single-database-check group, despite SQL Server otherwise supporting cross-database three-part names. This method is only invoked from the dbt-postgres and dbt-redshift macro packages — grepping the loader's macro assets turns up no other caller — and SQL Server's own macro package never dispatches into either.

Left alone

• adapter_specific_behavior_flags: returns vec![] for SqlServer. v1 registers 5 behavior flags, but 4 (dbt_sqlserver_use_default_schema_concat, dbt_sqlserver_use_native_string_types, dbt_sqlserver_enable_safe_type_expansion, dbt_sqlserver_use_dbt_transactions) stay on their v1 default with no alternate code path in the Rust adapter yet (e.g. Part 6 hardcoded the native-string-type mapping unconditionally, with no legacy branch reading the flag), and the 5th (dbt_sqlserver_disable_empty_relation_aliases) is out of scope here. None are registered rather than declaring toggles the adapter doesn't yet honor — worth a follow-up once/if those code paths land.

Verification

• cargo check -p dbt-adapter --tests: clean, 0 errors.
• cargo build -p dbt-adapter: clean.
• cargo test -p dbt-adapter --lib: 891 passed, 0 failed.
• cargo fmt: applied.

This closes out the full compile-error backlog for the dbt-adapter crate — Parts 4/5/6/7 combined leave it building clean end to end.

Closes #7