Live pod GPU capacity and per-class rates

What a pod create can actually claim right now, per GPU model: free native (device-request) GPU units on online, non-delisted nodes (the exact filter admission uses), plus the SERVER-computed hourly rate per compute class (interruptible is discounted below the on-demand/reserved rate). Only models with pod units and a price are returned; a create flow should offer exactly these and render these rates verbatim rather than re-deriving prices from the catalog.