chore(core): Normalization of total_cost for prompt key in llm_cache

[Mohammad Mohtashim (keenborder786)]

• Simple Normalisation of total_cost to align with converted_generation to hit the cache with the same message content if a message history is injected.
• Fixes: Cache hit total_cost injection breaks downstream cache keys for multi-turn conversations #35308

[codspeed-hq]

Merging this PR will improve performance by 35.8%

⚠️ Unknown Walltime execution environment detected

Using the Walltime instrument on standard Hosted Runners will lead to inconsistent data.

For the most accurate results, we recommend using CodSpeed Macro Runners: bare-metal machines fine-tuned for performance measurement consistency.

⚡ 12 improved benchmarks
✅ 1 untouched benchmark
⏩ 23 skipped benchmarks¹

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	WallTime	test_import_time[HumanMessage]	274.2 ms	245.6 ms	+11.66%
⚡	WallTime	test_import_time[PydanticOutputParser]	573.6 ms	485.9 ms	+18.03%
⚡	WallTime	test_import_time[RunnableLambda]	519.5 ms	450.1 ms	+15.42%
⚡	WallTime	test_import_time[InMemoryVectorStore]	625.8 ms	554.2 ms	+12.91%
⚡	WallTime	test_import_time[CallbackManager]	328.2 ms	290.4 ms	+13.04%
⚡	WallTime	test_import_time[ChatPromptTemplate]	660.6 ms	567 ms	+16.51%
⚡	WallTime	test_import_time[BaseChatModel]	560.9 ms	487 ms	+15.18%
⚡	WallTime	test_import_time[Runnable]	513.6 ms	447.7 ms	+14.72%
⚡	WallTime	test_async_callbacks_in_sync	25.8 ms	19 ms	+35.8%
⚡	WallTime	test_import_time[Document]	197.5 ms	170.7 ms	+15.66%
⚡	WallTime	test_import_time[LangChainTracer]	475 ms	420.6 ms	+12.93%
⚡	WallTime	test_import_time[tool]	561.9 ms	486.4 ms	+15.51%

---

_{Comparing keenborder786:mmk/store_cached_generation (d83eddd) with master (cdf140e)}

Footnotes

1. 23 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩