feat: refactor agent memory representation and modify routes for editing blocks

[sarahwooders]

Please describe the purpose of this pull request.
This PR aims to resolve bugs resulting from inconsistent agent state and blocks, as well as simplifying the agent creation REST API route.

Changes to agent state

This PR refactors the schemas related to the agent's state:

• PersistedAgentState: This is the schema defining what is persisted in the agents table
• CreateAgent: This is the schema defining the route for agent creation. This PR simplifies is so that memory is created by passing in a list of CreateBlock objects, rather than a Memory object
• AgentState: This schema contains all the state related to an agent contained in process memory, that is passed into Agent to instantiate an agent. It contains additional objects like the Memory object, sources, tools, and tags - since information for these is also contained in separate tables. We use this to both create the Agent class and also as the return on the client GET for agents.

Update block routes

This PR also cleans up the routes related to blocks:

• Blocks should primarily be edited via the block_id - the block routes are on /blocks/{block_id}
• Agent-block routes (e.g. attaching a block to an agent, editing an agen't block) are done via the /agents/blocks/{block_label} route, using the block_label as the reference

Removal of agent caching

We no longer cache agents - agent state must always be read from the DB, and the interface to pass messages to must be passed between functions.

Truncation of error messages

For tool error messages, we add a MAX_ERROR_MESSAGE_CHAR_LIMIT constant to limit the size of the error message passed into the context window.

Is your PR over 500 lines of code?
Yes

[mattzh72]

Some small bugs left:

FAILED tests/test_memory.py::test_dump_memory_as_json - AttributeError: 'ChatMemory' object has no attribute 'to_dict'
FAILED tests/test_memory.py::test_load_memory_from_json - AttributeError: 'ChatMemory' object has no attribute 'to_dict'
FAILED tests/test_memory.py::test_memory_jinja2_template_load - AttributeError: 'ChatMemory' object has no attribute 'to_dict'
FAILED tests/test_memory.py::test_memory_jinja2_template - AttributeError: 'ChatMemory' object has no attribute 'memory'
FAILED tests/test_memory.py::test_memory_jinja2_set_template - ValueError: Prompt template is not compatible with current memory structure: 'ChatMemory' object has no attribute 'memory'
FAILED tests/test_memory.py::test_link_unlink_block - AttributeError: 'ChatMemory' object has no attribute 'link_block'
FAILED tests/test_memory.py::test_update_block_label - AttributeError: 'ChatMemory' object has no attribute 'update_block_label'. Did you mean: 'update_block_value'?
FAILED tests/test_memory.py::test_update_block_limit - AttributeError: 'ChatMemory' object has no attribute 'update_block_limit'. Did you mean: 'update_block_value'?
FAILED tests/test_summarize.py::test_summarize_messages_inplace - pydantic_core._pydantic_core.ValidationError: 1 validation error for Memory
blocks.0
  Input should be a valid dictionary or object to extract fields from [type=model_attributes_type, input_value=None, input_type=NoneType]
    For further information visit https://errors.pydantic.dev/2.9/v/model_attributes_type
FAILED tests/test_summarize.py::test_auto_summarize - pydantic_core._pydantic_core.ValidationError: 1 validation error for Memory
blocks.0
  Input should be a valid dictionary or object to extract fields from [type=model_attributes_type, input_value=None, input_type=NoneType]
    For further information visit https://errors.pydantic.dev/2.9/v/model_attributes_type

[mattzh72]

overall looks good, some small minor code quality nits, but will probably get cleaned up over time.