Fix HTTP transport message handling

[pavelanni]

I started testing the HTTP transport using the provided server and client in examples/http_example. It didn't work for me, and I asked Cursor/Sonnet to help me with that. After adding logging to different parts of the code, we (of course, mostly Claude/Sonnet, not me) finally found the issue and fixed it. Now, the client works as expected and displays time in different formats. Here is the explanation and the changes:

Issue

The HTTP transport implementation had several issues that prevented proper message handling:

1. Message handlers were not being properly set on the base transport
2. Response channels were managed in two different locations (HTTPTransport and baseTransport)
3. Redundant state was maintained between the transport layers

Changes

1. Fixed SetMessageHandler to properly set the handler on the base transport while maintaining backward compatibility
2. Removed redundant responseMap field from HTTPTransport struct
3. Removed redundant responseMap initialization from NewHTTPTransport
4. Updated Send method to use the correct responseMap from base transport

Impact

• HTTP transport now correctly handles JSON-RPC messages and responses
• Simplified code by removing redundant state
• Fixed message handler delegation between transport layers

Testing

The changes have been tested with JSON-RPC initialize requests and confirm that:

• Message handlers are properly called
• Responses are correctly sent back to clients
• No messages are lost between transport layers

[rvoh-emccaleb]

Possible duplicate of #66, but that one is temporarily closed

[pavelanni]

Possible duplicate of #66, but that one is temporarily closed

Yes, it looks like it's the same issue. Is there a plan to merge the fix (either this one or #66)?

[rvoh-emccaleb]

@pavelanni -- I heard from @Chrisbattarbee, here, three weeks ago when I first opened #66. But I haven't heard since.

Sounded like he was crunched for time, then. Been running off my fork since.

[a67793581]

Yes, it looks like it's the same issue. Is there a plan to merge the fix (either this one or #77)?