TestSprite 2.1 speeds testing, guards PRs

This is the kind of update that makes AI testing feel less like a demo and more like real delivery infrastructure.

• –The biggest win is workflow fit: TestSprite now sits inside MCP-driven coding loops locally, then enforces the same checks in GitHub on every PR
• –Cutting test execution to about five minutes matters because slow verification kills the speed gains promised by AI coding tools
• –The visual test editor lowers the maintenance tax that usually makes autogenerated end-to-end tests brittle and annoying to keep
• –Support for live preview deployments and merge blocking pushes the product from “AI writes tests” into “AI can actually protect production quality”
• –TestSprite is clearly aiming at the same AI-native stack as Cursor, Claude Code, and Copilot users who need stronger feedback loops than code generation alone provides