fix: enforce 255-character identifier length limit for Databricks relations

[psaikaushik]

Summary

Enforces the 255-character identifier length limit for Databricks relations at relation creation time, preventing a cryptic runtime DatabricksExecutionError when store_failures generates overly-long table names.

Closes #1309

Problem

When store_failures: true is enabled, dbt generates table names by concatenating the test name, model name, and arguments. For deep schema tests with verbose names, this can exceed Databricks' 255-character limit for table names, causing:

[RequestId=...] Invalid input: RPC CreateStagingTable Field
managedcatalog.StagingTableInfo.name: name "accepted_map_keys_stg_product_search_create_sizes__relax__..." too long.
Maximum length is 255 characters.

This error is unhelpful — it doesn't explain what the 255-character limit is, which test/model caused it, or how to fix it.

Fix

Added relation_max_name_length() and __post_init__ validation to DatabricksRelation, following the exact same pattern used by the Postgres adapter:

MAX_CHARACTERS_IN_IDENTIFIER = 255

def relation_max_name_length(self):
    return MAX_CHARACTERS_IN_IDENTIFIER

def __post_init__(self):
    if (
        self.identifier is not None
        and self.type is not None
        and len(self.identifier) > self.relation_max_name_length()
    ):
        raise DbtRuntimeError(
            f"Relation name '{self.identifier}' "
            f"is longer than {self.relation_max_name_length()} characters. ..."
        )

Before:

[RequestId=...] Invalid input: RPC CreateStagingTable ... name "accepted_map_keys_stg_..." too long.

After:

Relation name 'accepted_map_keys_stg_product_search_create_sizes__...'
is longer than 255 characters. Databricks has a maximum identifier
length of 255 characters. If this is a store_failures table, consider
using a shorter test name or setting a custom alias.

Checklist

• Follows the same pattern as the Postgres adapter
• Catches the error at relation creation time, not at SQL execution
• Error message includes actionable guidance