Sergi Siso

I updated the issue title and description with the current inlining stats for gungho_model LFRic miniapp

This is also one of the major issues in OpenMP for GPU now, mostly in the sbcblk region. Some functions that will need to be inlined are: q_sat, sbc_dcy, gamma_moist,...

Alternatively, now that we have the elemental attribute, we can a resolve the `cp_air` symbol (which could be imported) for: `my_array(:,:) = cp_air(array1(:,:)) * array2(:,:) + array3(:,:)` and then arrayrange2loop...

I agree with the `get_` but maybe it could be `Routine.Symbol.get_routine()` instead of "schedule" to make it easier for people not familiar with the PSyIR concepts. (PS: Now I would...

I am checking with everybody if we all agree that the NEMO API doesn't need a name and psyclone without an API will just do code transformation without any assumption...

The resulting shape of a(3:9:3) `node.datatype.shape` is indeed "a 1D array of extent (9-3)/3 + 1 which implicitly starts at index 1." You can get the lower bound of "a"...

or forbid inlining such routines for now.

Regarding the last comments on this issue, in #1973 @arporter has introduced querying the array lower bound in ArrayMixin that choses between `node.symbol.datatype.shape.lower` or an `LBOUND` expression depending on the...

A `datatype` property on all DataNodes and support for operations was added with #2387. This means that we can start to use/depend on this functionality. But we still need better...

> respect to last time I tried (master?) Not master, this PR makes 2 more calls able to be inlined compared to master, this means that almost all the interfaces...