fix(storage): show human-readable error for uninitialized sqlite data…

[TryingtobeingNikhil]

What problem does this fix?

This fixes Issue #469 where launching the dashboard against an entirely empty or uninitialized SQLite database crashes the application with a confusing SQLAlchemy backend traceback rather than providing guidance to the user.

Root cause

When RDBStorage initializes with skip_table_creation=True, it throws an underlying sqlite3.OperationalError: no such table inside a generic wrapper exception because the Optuna schema does not exist yet.

What the fix does

It cleanly intercepts optuna.exceptions.StorageInternalError during get_rdb_storage() initialization, specifically inspects the __cause__ string for missing table warnings, and prints clear CLI instructions on how to initialize a database before exiting with code 1.

How I tested it

• Manually ran optuna-dashboard sqlite:///empty_db.sqlite3 on a blank file and confirmed the new CLI instruction message appeared and exited immediately.
• Manually generated a valid study database via a Python script and verified the dashboard booted correctly without hitting the new error trap.
• Added a new pytest unit test simulating the empty-database crash and asserting the SystemExit code 1 behavior.
• Ran the entire python_tests/ suite locally to ensure no other URL inference testing logic broke.

Edge cases considered

• Inspected e.__cause__ because the true missing table detail gets wrapped deeply inside the SDK's generic exception structure.
• Re-raised the unchanged StorageInternalError if it is triggered by an unrelated issue (e.g. data length violation) to avoid accidentally swallowing real bugs.