Provide LLM reasoning effort param

[zfoong]

What and why:
Many LLM providers take reasoning effort or thinking budget as params. When none are supplied, the default can be None or medium. This hurts CraftBot's performance, causing suboptimal performance like ending tasks prematurely, ignoring obvious context, and producing bad results and output. This PR supplies the LLM call with these parameters.

However, each provider, and even models within the same provider, can take in different parameters. Therefore, the rules are ALL hard-coded in models/reasoning class. The reason is simple: there is no way to obtain them dynamically.

This can cause a potential issue where any update to the API call will make the list obsolete.

[zfoong]

@ahmad-ajmal Updated! Request review again.