Add fallback coverage for unsupported Paimon scan data type

[weimingdiit]

Is your feature request related to a problem? Please describe?

Paimon native scan rejects columns whose data types are not supported by Auron.
The fallback path is implemented, but the current integration suite does not verify
that unsupported scan data types remain on Spark execution.

Without this coverage, future changes to Paimon scan conversion could accidentally
allow unsupported types into the native path.

Describe the solution you'd like

Add integration coverage for a Paimon table containing an unsupported decimal type.
The test should verify that:

• Spark still returns the correct result.
• The query falls back from NativePaimonV2TableScan.

Describe alternatives you've considered

No execution logic changes are proposed. This change only adds regression coverage
for the existing fallback behavior.

Additional context

The test uses decimal(38,10), which is outside Auron's currently supported native
decimal range.