Add asyncio support for LLM (OpenAI), Chain (LLMChain, LLMMathChain), and Agent (#841)

Supporting asyncio in langchain primitives allows for users to run them
concurrently and creates more seamless integration with
asyncio-supported frameworks (FastAPI, etc.)

Summary of changes:

**LLM**
* Add `agenerate` and `_agenerate`
* Implement in OpenAI by leveraging `client.Completions.acreate`

**Chain**
* Add `arun`, `acall`, `_acall`
* Implement them in `LLMChain` and `LLMMathChain` for now

**Agent**
* Refactor and leverage async chain and llm methods
* Add ability for `Tools` to contain async coroutine
* Implement async SerpaPI `arun`

Create demo notebook.

Open questions:
* Should all the async stuff go in separate classes? I've seen both
patterns (keeping the same class and having async and sync methods vs.
having class separation)

tests/integration_tests/llms/test_openai.py

@@ -7,6 +7,7 @@ import pytest
 
 from langchain.llms.loading import load_llm
 from langchain.llms.openai import OpenAI
+from langchain.schema import LLMResult
 
 
 def test_openai_call() -> None:
@@ -74,3 +75,11 @@ def test_openai_streaming_error() -> None:
     llm = OpenAI(best_of=2)
     with pytest.raises(ValueError):
         llm.stream("I'm Pickle Rick")
+
+
+@pytest.mark.asyncio
+async def test_openai_async_generate() -> None:
+    """Test async generation."""
+    llm = OpenAI(max_tokens=10)
+    output = await llm.agenerate(["Hello, how are you?"])
+    assert isinstance(output, LLMResult)

tests/unit_tests/chains/test_hyde.py

@@ -33,6 +33,11 @@ class FakeLLM(BaseLLM, BaseModel):
     ) -> LLMResult:
         return LLMResult(generations=[[Generation(text="foo") for _ in range(self.n)]])
 
+    async def _agenerate(
+        self, prompts: List[str], stop: Optional[List[str]] = None
+    ) -> LLMResult:
+        return LLMResult(generations=[[Generation(text="foo") for _ in range(self.n)]])
+
     @property
     def _llm_type(self) -> str:
         """Return type of llm."""