Support for Reading Secrets from Files via Mounted Volumes

[ChrisJBurns]

Injecting secrets into containers as environment variables is a common approach many take. However, it isn't the most secure due to increased change of leaking. We should support the ability to read secrets as files within ToolHive and the Operator. The underlying MCP servers would already have this as an option because we offer the podTemplateSpec override in the MCPServer CRD. But for ToolHive and the Operator itself, we should consider supporting it also.

As we do not need it for now, this can be a lower priority.