docs: Fix Guardrail intro to include output validation

[Wania-Kazmi]

Problem:
The original introductory sentence stated that guardrails enable validation for just "user input".

Solution:
This change updates the sentence to clarify that guardrails enable validation of "data flowing through your agent system (such as user input or agent output)".

Why this change?
This provides a more accurate and comprehensive initial understanding of what guardrails are used for, as the documentation later details both Input Guardrails (for user input) and Output Guardrails (for agent output). Aligning the introduction makes the documentation clearer from the start.

Specific Change:

• Old: Guardrails run in parallel to your agents, enabling you to do checks and validations of user input.
• New: Guardrails run in parallel to your agents, enabling you to perform checks and validations on data flowing through your agent system (such as user input or agent output).