Update report: generalization stress-test, planner ceiling, two-tier solution

Add agentic_demo.py: agentic system report + proven fast-mode demo

Add report.md: agentic system report + proven fast-mode demo

Update app.py

Update README.md

Unlimited output (max_tokens=-1); raise default N_CTX to 32768

Fix: pin threads to 2 (os.cpu_count oversubscribed cpu-basic)

Add enable_thinking chat template, auto threads; true fast mode

Serve copied Qwopus3.5-0.8B (anon334test/qwopus) via llama-server, based on testgguf/codingagent

initial commit