[ENH] Improve Documentation for `pytorch-forecasting` (both v1 and v2)

[phoeenniixx]

Is your feature request related to a problem? Please describe.
Currently the pytorch-forecasting documentation is cluttered and have no clear distinction between v1 and v2 APIs. Also, we still dont have references of new models that have been added, and the UX is also not good

Describe the solution you'd like
Improve the documentation-

for v1

• make the models table automated and add new models which are not present - [ENH] auto-generate model overview table from registry tags #2236
• make the metrics table less cluttered and also automate it to be created via tags

for v2

• Add some references for v2 to readme as well
• Create a new header for v2 API
• make the models table automated and add new models which are not present.
• Add separate tutorials for v2 - should be easily distinguishable from v1
• Add API references
• Extension Templates
• separate v2 models and datamodules from v1
• Clearly mention that metrics are still the same as v1

[phoeenniixx]

FYI @agobbifbk, @fkiraly, @PranavBhatP

[Soham-47]

hey @phoeenniixx , could I take this up?

[phoeenniixx]

Hi @Soham-47
Yes, ofcourse!
For now, I would suggest improving metrics table and generate it directly from the tags and metric_pkg, similar to what @IgnazioDS is doing in #2236

[PranavBhatP]

Can we have something like sktime-mcp for ptf @phoeenniixx ?

[phoeenniixx]

I think this is a good idea to have an mcp for ptf, but imo, we first need to get v2 and FM integration and make the API stable....
FYI @fkiraly @agobbifbk