Feat/raae 1094 add svs vamana implementation

[nkanu17]

Adds first-class SVS-VAMANA support to redis-vl-python, including compression options, schema validation, Redis capability checks, docs, and tests.

Highlights

• New SVSVectorField + VectorIndexAlgorithm.SVS_VAMANA.
• Compression enums: LVQ4, LVQ4x4, LVQ4x8, LVQ8, LeanVec4x8, LeanVec8x8.
• Validation: only float16/float32; reduce < dims; reduce allowed only with LeanVec.
• Capability checks (sync/async): block create if requirements not met; clear RedisModuleVersionError.
• CompressionAdvisor: recommend configs + estimate memory savings.
• Parsing: normalize SVS attrs (incl. basic compression name mapping).

Requirements

• Redis ≥ 8.2.0 and RediSearch ≥ 2.8.10 (or SearchLight ≥ 2.8.10) for SVS-Vamana
• Backwards compatible, just can't use SVS-Vamana.

Compatibility

• No breaking changes for flat/hnsw.
• Backwards compatible, just can't use SVS-Vamana.

Docs & Tests

• Docs: schema.rst adds SVS section + example.
• Tests: validation, capability detection, CompressionAdvisor, and basic integration.
• Notebook or python guide on usage

[rbs333]

This is good! I think let's add a recipe to the user guide and then we can talk more about the migration class in our 1:1

[abrookins]

Nice! I think we should talk about migrations, but agree that in this case, for migrating a specific aspect of an index rather than more general schema migrations support, or data migrations, a recipe may be best.

[rbs333]

Generally looks really good!

Left a few comments for tweaks but otherwise good. After the tweaks please post some screen shots of the new documentation changes in #applied-ai to see that other members of the team are good with them.