[WIP][python] Route reads to io-cache targets per file type

[sundapeng]

Purpose

The pypaimon side of #10350: when a REST table token vends OSS cache endpoints, RESTTokenFileIO wraps the OSS FileIO so that immutable files are read through a cache endpoint chosen by file type, and every other request stays on the OSS endpoint. The options and rules are the same as in #10350:

• io-cache.endpoint names one cache endpoint; io-cache.targets with io-cache.target.<name>.endpoint (and optional .region, .path-style-access) names several, and io-cache.routes maps file types to them. io-cache.policy, io-cache.whitelist and io-cache.origin.endpoint work as in Java, and io-cache.enabled set in the catalog options overrides the token.
• Only reads and file status use a cache. Writes, exists, listings, deletes, renames, copies and non-oss:// paths stay on OSS; a client-side dlf.oss-endpoint turns routing off.
• Files rewritten in place (FileType.is_mutable) and files named by a sequence id never use a cache.
• Data files are recognized by name: a data or changelog file prefix, including a custom one, followed by {uuid}-{count}.{extension}. Manifests, indexes and statistics have no extension, and the manifest sidecar is excluded explicitly. Names Paimon does not write stay on OSS.
• Errors from a cache endpoint are raised as they are, without retrying on OSS.
• The PyArrow, Lance and Vortex readers use the endpoint, scheme and addressing style of the chosen target (fs.oss.second.level.domain.enable for path-style).

The test cases in tests/resources/io_cache are the same files as in #10350 and apache/paimon-rust#1014.

Tests

• io_cache_routing_test.py runs the shared cases (82 request cases, 64 file-name cases) and the option parsing.
• io_cache_file_io_test.py: which operations use a target, per-target file systems, errors from a target.
• rest_token_file_io_test.py, file_io_test.py and lance_utils_test.py cover the token merge, the PyArrow endpoint and addressing, and the Lance and Vortex storage options.