Fine tuning docs

[kavanase]

Adds some updates to the docs on fine-tuning, namely:

• Cross-linking the in-depth fine-tuning discussion in the mini-section on fine-tuning in the workflow page
• Add a section on 'Considerations for Fine-Tuning Training', to just note the main things that users should consider when fine-tuning (which differ from from-scratch training). This was motivated by my discussions with a few different potential fine-tuning users, who were asking what the main things they should change vs from-scratch are, so seemed useful to just state the main points (drop LR, update shifts, note that model config is fixed).

If there is anything else useful to add, worth noting here. This section could possibly be updated in the future with more feedback and experience