[FIP-28] Support point lookup for historical partitions

[luoyuxia]

Search before asking

• I searched in the issues and found nothing similar.

Description

Parent issue: #3631

Support point lookup for historical partitions of datalake-enabled primary key tables. When a lookup targets an expired Fluss partition whose data has been tiered to lake storage, the client/server path should resolve the historical partition and read the matching row from the lake table instead of failing with partition-not-found behavior.

The implementation should cover:

• Client lookup request plumbing for carrying the original partition name when the active Fluss partition no longer exists.
• Server-side routing that distinguishes normal lookup from historical partition lookup.
• A lake lookup SPI and Paimon implementation for point lookup by primary key.
• Schema evolution behavior when old lake data is returned through the current Fluss table schema.
• Flow control for historical lake lookup so it cannot exhaust normal lookup resources.
• Unit and integration coverage for normal lookup, historical lookup, mixed routing, Paimon lookup, and schema evolution.

Related PR: #3630

Willingness to contribute

• I'm willing to submit a PR!