chore(deepinfra): add 45 new models and update pricing

[RioPlay]

Summary

Added 45 new model definitions and updated pricing across all DeepInfra models to match current deepinfra.com listings.

Models Added (45)

DeepSeek (7): R1-0528-Turbo, R1-Distill-Llama-70B, V3, V3-0324, V3.1, V3.1-Terminus, V3.2, V4-Flash
Qwen (17): 3-14B, 3-30B-A3B, 3-32B, 3-Coder-480B-Instruct, 3-Next-80B-A3B, 3-VL-30B-A3B, 3-VL-235B-A22B, 3-235B-A22B-Instruct-2507, 3-235B-A22B-Thinking-2507, 3.5 (0.8B, 2B, 4B, 9B, 27B, 35B-A3B, 122B-A10B, 397B-A17B), 3.6 (27B, 35B-A3B)
Google (5): gemma-3 (4B, 12B, 27B), gemma-4 (26B-A4B, 31B)
NVIDIA (5): Llama-3.3-Nemotron-Super-49B-v1.5, Nemotron-3-Nano-30B-A3B, Nemotron-3-Nano-Omni-30B-A3B-Reasoning, NVIDIA-Nemotron-3-Super-120B-A12B, NVIDIA-Nemotron-Nano-9B-v2
Meta (2): Llama-3.2-11B-Vision-Instruct, Llama-Guard-4-12B
Mistral (3): Mistral-Nemo-Instruct-2407, Mistral-Small-24B-Instruct-2501, Mistral-Small-3.2-24B-Instruct-2506
Moonshot (3): Kimi-K2.5, Kimi-K2.6, Kimi-K2-Instruct-0905
GLM (2 new files): GLM-4.5, GLM-4.6, GLM-4.6V, GLM-4.7, GLM-4.7-Flash, GLM-5, GLM-5.1
Others (5): phi-4, Step-3.5-Flash, Hermes-3-Llama-3.1-405B, Hermes-3-Llama-3.1-70B, MythoMax-L2-13b, L3-8B-Lunaris-v1-Turbo, L3.1-70B-Euryale-v2.2, gpt-oss-120b, gpt-oss-120b-Turbo, gpt-oss-20b

Pricing Updates

• GLM-5.1: input 1.4→1.05, output 4.4→3.50, cache 0.26→0.205
• GLM-4.7: input 0.43→0.40
• GLM-5: input 0.8→0.60, output 2.56→2.08, cache 0.16→0.12
• MiniMax-M2.5: input 0.27→0.15, output 0.95→1.15
• MiniMax-M2: updated input/output
• Kimi-K2.5: input 0.50→0.45, output 2.80→2.25, cache 0.09→0.07
• Qwen3.5-35B-A3B: input 0.2→0.18, output 0.95→1.00
• Qwen3.6-35B-A3B: input 0.20→0.15, output 1.00→0.95
• Qwen3-Coder-480B-Turbo: input 0.3→0.30, output 1.2→1.00
• gpt-oss-120b: input 0.05→0.039, output 0.24→0.19

Fixes

• Renamed invalid cache fields (cached, cached_input, cached_read) → cache_read across 4 files (deepinfra)
• Added missing cache_read for GLM-4.7-Flash and MiniMax-M2.5
• Added missing # https://deepinfra.com/... comment headers
• Fixed trailing whitespace and missing trailing newlines on all files

NOTE:
** AI was used to complete this task **

[RioPlay]

Only reason I reverted the vercel item was because it's out of scope for this particular submission.