Add response duration tracking to webhook message attempts

[CodeMan62]

Motivation

Currently, the OSS version of Svix doesn't record webhook response times, which is an important metric for monitoring webhook delivery performance. This capability exists in the commercial API (as documented at https://api.svix.com/docs#tag/Message-Attempt/operation/v1.message-attempt.get), but was missing in the open-source version.
Adding response duration tracking will provide OSS users with valuable information about webhook latency, helping them troubleshoot delivery issues and monitor the performance of their webhook endpoints.

Solution

This PR adds webhook response duration tracking to the OSS version of Svix by:

1. Adding a response_duration_ms field to the messageattempt database model
2. Creating a database migration to add this column to existing tables
3. Updating the webhook client code to measure and store response times using std::time::Instant
4. Modifying the MessageAttemptOut implementation to use the actual duration instead of hardcoding to 0
5. Removing documentation comments that mentioned response times not being available in OSS
   With these changes, the response time in milliseconds will be available through the API for each webhook attempt, providing parity with the commercial API's functionality.

[svix-jplatte]

Thanks for the PR!

For formatting, you might have to install a nightly toolchain and run rustfmt throught that (rustup install nightly + cargo +nightly fmt if you have rustup).

For the other failing CI check, simply run the generate-openapi.sh script in the server directory and the OpenAPI spec file will be updated.

[CodeMan62]

Hey @svix-jplatte, thanks for the review. I have done all the things you said. Take a look at the changes and tell me if any changes are required

[svix-jplatte]

Looks good to me!