Improve vectorstore query validation and error context

[deepintai]

Context

The vectorstore abstraction supports several query operators (Equal, GreaterThan, Like, ContainsAny, etc.) across backends (Redis, Weaviate, Qdrant, Pinecone). When a query uses an operator a backend does not support, the failure surfaces as the generic ErrNotSupported with no indication of which field or operator caused it — hard to diagnose.

What to do

1. Add a pre-flight validator in store.go:

   func ValidateQuery(query Query, supportedOperators []QueryOperator) error

   that rejects operators not in the supported set and surfaces type mismatches.
2. Extend errors.go with ErrInvalidOperator and ErrInvalidQueryValue sentinel errors.
3. Add a helper SupportedOperatorsForBackend(backendType VectorStoreType) []QueryOperator.
4. In at least one backend (e.g. redis.go), wrap query failures with field name and operator, e.g. "query failed on field 'age' with operator 'Like': operation not supported".

Acceptance criteria

• ValidateQuery can be called before execution and returns errors that name the offending field and operator.
• New sentinel errors added and used.
• At least one backend produces enhanced, field/operator-aware error messages.
• Errors distinguish "unsupported operator" from a genuine backend query failure.
• New unit tests assert the error messages; go test ./framework/vectorstore/... passes.

File pointers

• Store interface + operators: deepintshield_server/framework/vectorstore/store.go (interface ~80-107, operators ~28-42)
• Errors: deepintshield_server/framework/vectorstore/errors.go
• Example backend: deepintshield_server/framework/vectorstore/redis.go