Support shared/linked buckets from other projects in direct Storage access

[beckavac]

Problem

When planning to remove input mapping from a Snowflake transformation and replace table aliases with direct Keboola Storage paths (format: sapi_<project_id>."<bucket-id>"."<table-name>"), shared buckets sourced from other projects are not visible in the current project's Snowflake database.

Concrete example

Project Slevomat (ID: 226) has bucket in.c-db which is a shared bucket from a different project (not owned by project 226). When running SHOW SCHEMAS IN DATABASE sapi_226, the schema in.c-db does not appear — even though it is accessible via input mapping.

This means the direct-access path sapi_226."in.c-db"."orders" may not work, and the correct path likely points to a different Snowflake database (the source project's database).

Expected behavior

kbagent should be able to:

1. Detect shared/linked buckets — when listing input tables or analyzing a transformation's input mapping, flag which buckets are shared from other projects vs. owned locally.
2. Resolve the correct Snowflake path for shared buckets — return the full qualified path including the source project's database (e.g., sapi_<source_project_id>."<bucket-id>"."<table-name>").
3. Ideally expose this via a command like:

kbagent --json bucket info --project slevomat --bucket in.c-db
# Should return: owner project, Snowflake database, direct access path

Use case

This is needed for the optimization pattern of removing input mapping from large transformations to eliminate the data-copy overhead (in this case: 17 minutes for 209 tables). Without knowing the correct Snowflake path for shared buckets, the migration cannot be completed safely.

Context

• Project: Slevomat (ID: 226), stack: connection.keboola.com
• Transformation: Slevomat Main Task (config ID: 746772863)
• Affected bucket: in.c-db — 161 tables (77% of all input tables in the transformation)
• The bucket appears in input mapping as in.c-db.* but its schema is absent from SHOW SCHEMAS IN DATABASE sapi_226

Proposed solution

• Add a bucket info subcommand (or extend config detail) to resolve the Snowflake-level path for each input table, including cross-project shared buckets
• In config detail output, add field snowflake_path alongside each input table entry

[padak]

Partially addressed in PR #46 -- added Snowflake quoting best practice to kbagent context (tip #15):

ALWAYS double-quote database, schema, and table names in Snowflake workspace queries. Unquoted identifiers become UPPERCASE (e.g., sap_9 -> SAP_9 -> not found).

-- WRONG:
SELECT * FROM sapi_226.in.c-db.orders
-- RIGHT:
SELECT * FROM "sapi_226"."in.c-db"."orders"

This solves the quoting dimension of the problem. The full shared bucket path resolution (detecting source project DB for cross-project buckets) remains open.

[padak]

Implemented in PR #47 -- new kbagent storage bucket-detail command resolves Snowflake paths for shared/linked buckets. For in.c-db linked from project 1507, it correctly returns "sapi_1507"."in.c-keboola-ex-db-mysql"."table_name" instead of the wrong sapi_226 path. Thanks for the detailed issue description!

[padak]

Resolved in PR #47. New kbagent storage bucket-detail command resolves Snowflake paths for shared/linked buckets. Also added Snowflake quoting best practice to kbagent context.