Retry functionality for OpenAPI skills

[SergeyMenshykh]

Motivation and Context

This PR adds two improvements:

• It extends all ImportOpenAPISkill* and ImportChatGptPluginSkill* methods to accept HTTP retry configuration.
• It improves HttpClient usage in OpenAPI skills by reusing the same instance for the same set of imported skills.

The first improvement will allow OpenAPI skill to be more resilient against unreliable REST API it tries to access.

The second improvement will decrease number of HttpClient instances OpenAPI skills create. Instead of creating one instance of HttpClient per skill, one instance will be created per set of skills that are imported together by the same Import* method. This is the first step to decrease number of HttpClient created for OpenAPI skills and it's not the final one. Later, when agreed on a proper solution, the HttpClient could be externalized so that client code/hosting app could provide its own instance, if required.

Description

1. All the ImportOpenAPISkill* and ImportChatGptPluginSkill* methods are extended to accept and respect HttpRetryConfig.
2. Functionality that creates HttpClient is moved one level up so that it's called once per import rather than for each skill in the import.

Contribution Checklist

• The code builds clean without any errors or warnings
• The PR follows SK Contribution Guidelines (https://github.com/microsoft/semantic-kernel/blob/main/CONTRIBUTING.md)
• The code follows the .NET coding conventions (https://learn.microsoft.com/dotnet/csharp/fundamentals/coding-style/coding-conventions) verified with dotnet format
• All unit tests pass, and I have added new tests where possible
• I didn't break anyone 😄