PhariaAI v1.251000.2

Fixed an issue where adding filters to search queries did not work on search stores without indexes.

Fixed a permission bug on empty search stores that resulted in a 500 error in PhariaAssistant.

Fixed an issue where spaces in chat messages were sometimes not emitted.

Fixed a problem where database connections were not closing properly, leading to increased latency.

Fixed an issue where selected documents, collections, and tools disappeared after the first message was sent.

Improved handling of "thinking" behavior before final answer generation in express mode.

Improved proxy liveness probe in inference-api-proxy.

Improved worker readiness probe stability in inference-worker.

Updated inference-worker to new vLLM version 0.11.

Improved backpressure handling in inference-api by forwarding rate limit status 408 instead of an internal error.

Updated PhariaCatch UI and API dependencies.

Improved error logging in document-index.

Added support for strict guardrails and custom guardrail rejection messages for PhariaAssistant.