feat(toolbox-langchain)!: Base toolbox-langchain over toolbox-core

[anubhav756]

This PR introduces a new dependency on the foundational library toolbox-core.

Breaking Changes

This PR introduces changes to how "strictness" is handled across several toolbox functions. The strict flag has been removed from load_tool, aload_tool, add_auth_token_getter(s), and bind_param(s). Additionally, the default behavior of load_toolset for strict=False has changed.

Why?

Enforces more explicit error handling and reduce ambiguity, particularly in scenarios where unused configurations might otherwise be silently ignored.

* (a)load_tool do not support strict flag now:

Before

tool = await client.aload_tool("my-tool", strict=True)

After

tool = await client.aload_tool("my-tool")
tool = await client.aload_tool("my-tool", strict=True) # <-- Error

* (a)load_toolset functionality changes for strict=False

• Before

tools = toolbox.load_toolset(
    auth_token_getters={...},
    bound_params={...},
    strict=False,
)

Does not throw error if none of the loaded tools use a given auth token getter or bound param. Only throws a warning.

After

tools = toolbox.load_toolset(
    auth_token_getters={...},
    bound_params={...},
    # strict=False, <-- default
)

Throws error if none of the loaded tools use a given auth token getter or bound param.

* add_auth_token_getter(s) do not support strict flag

Before

# Warning
auth_tool = tool.add_auth_token_getters(
    {"unused_auth": ...},
    strict=False,
)

# Error
auth_tool = tool.add_auth_token_getters(
    {"unused_auth": ...},
    # strict=True, <-- default
)

After

# Error
auth_tool = tool.add_auth_token_getters(
    {"unused_auth": ...},
)

* bind_param(s) do not support strict flag

Before

# Warning
bound_tool = tool.bind_params(
    {"unused_param": ...},
    strict=False,
)

# Error
bound_tool = tool.bind_params(
    {"unused_param": ...},
    # strict=True, <-- default
)

After

# Error
bound_tool = tool.bind_params(
    {"unused_param": ...},
)

[kurtisvg]

For breaking changes, let's make sure we call out what steps are required to migrate before hand. It's not clear to me which API is breaking here.

Ideally, we want this to be a non-breaking change. Is that possible?