fix(workflow): increase timeouts for workflow prompts

What does this merge request do and why?

Currently, we enforce 30 seconds timeout for Anthropic Python SDK client.

There's some specifics in Anthropic API server behavior: it stops streaming at some point for large input/outputs.

However, it sends a ping message every 30 seconds. As a result, we have a race condition and sometimes the ping message is not processed within the 30 seconds timeout window, so we get a timeout after 60 or 90 or 120 seconds. Let's increase the timeout window to 60 seconds. In this case, we'll have enough time to process the ping message if it's sent by the Anthropic server.

This commit removes the timeout overrides and uses the default value for Anthropic client: 60 seconds.

Related issues:

• Customer issue: https://gitlab.com/gitlab-com/account-management/emea/natwest/natwest-dedicated-collaboration/-/work_items/549#note_2753462675
• Issue related to the problem: LLM times out when serving a prompt with a lot ... (#1465 - closed)
• Investigations: #1465 (comment 2753349242)