fix(snowflake): accept column list after CREATE DYNAMIC TABLE options

[lustefaniak]

Summary

Snowflake permits the column-definition list to appear after the options header (TARGET_LAG / WAREHOUSE / REFRESH_MODE / INITIALIZE) rather than before it. This form is what generator-emitted dynamic-table DDL uses, and it previously failed to parse with Expected end of statement, found: <TYPE>.

CREATE OR REPLACE DYNAMIC TABLE t
    TARGET_LAG = '100000 Days'
    WAREHOUSE = COMPUTE_WH
    REFRESH_MODE = AUTO
    INITIALIZE = ON_CREATE
    ( "SYSTEM_CREATE_DATE" TIMESTAMP_NTZ(9), "ID" VARCHAR(256), ... )
    AS SELECT ...

Root cause

Two issues, both in parse_create_table:

• The bare-word option value (e.g. INITIALIZE = ON_CREATE) was parsed with parse_expr, which greedily treated the trailing column-list ( as a function call on the value word — swallowing ON_CREATE ( "col" TYPE, ... ). Bare-word option values are now parsed as a plain identifier so the ( is left intact.
• Nothing picked up a column list that appears after the options loop. The up-front parse_columns() runs before the options and saw the first option keyword, not (. A new arm in the options loop parses the column list when it follows the options.

Coverage

New tests cover: column list + AS SELECT, column-list-only (no AS), the no-whitespace "col"TYPE form, and the types TIMESTAMP_NTZ / TIMESTAMP_TZ / NUMBER / VARCHAR / DATE / BOOLEAN. Full suite green (1187 tests); corpus run shows 0 regressions.

https://claude.ai/code/session_012hHZLhjkLPneyPz7jpMWFS

[github-actions]

Corpus Parsing Report

Total: 191713 passed, 1920 failed (99.0% pass rate)

✨ No changes in test results

By Dialect

Dialect	Passed	Failed	Total	Pass Rate	Delta
ansi	511	69	580	88.1%	-
athena	37	1	38	97.4%	-
bigquery	42324	113	42437	99.7%	-
clickhouse	2723	129	2852	95.5%	-
databricks	2965	184	3149	94.2%	-
doris	28	12	40	70.0%	-
dremio	27	0	27	100.0%	-
duckdb	1124	45	1169	96.2%	-
exasol	54	7	61	88.5%	-
fabric	6	0	6	100.0%	-
generic	17	38	55	30.9%	-
hive	35	10	45	77.8%	-
materialize	6	14	20	30.0%	-
mssql	2276	407	2683	84.8%	-
mysql	151	37	188	80.3%	-
oracle	1046	352	1398	74.8%	-
postgres	1172	111	1283	91.3%	-
presto	55	8	63	87.3%	-
redshift	40503	65	40568	99.8%	-
singlestore	141	9	150	94.0%	-
snowflake	94741	138	94879	99.9%	+2
spark	90	20	110	81.8%	-
sqlite	51	16	67	76.1%	-
starrocks	29	4	33	87.9%	-
teradata	23	20	43	53.5%	-
trino	1413	77	1490	94.8%	-
tsql	165	34	199	82.9%	-