Dimension 0 in FLORIS v4

[misi9170]

In FLORIS v3, generally numpy arrays in the simulation module are 5D, with indices
(wd, ws, turbine, rotor_grid_index1, rotor_grid_index2). In time_series mode, the ws dimension is functionally collapsed to length 1.

FLORIS v4 will always operate in such a time_series mode (with the time_series=True option no longer explicit but simply the default behavior), and the length 1 dimension will be removed, resulting in 4D numpy arrays with indices (findex, turbine, rotor_grid_index1, rotor_grid_index2).

@rafmudaf @paulf81 and I discussed what the new collapsed zeroth index should be called. There are two issues here:

1. What the zeroth index variable should be referred to in the low-level source code
2. How we describe the zeroth dimension at the high level, in examples and documentation in particular.

1. Source code variable

For the first, we considered the following options:

• A physical descriptor, such as sample, condition, state, event
• A software descriptor, such as row, index, run, calculation, simulation, iteration, instance

We decided that physical descriptors are potentially subject to misinterpretation (as they may mean different things to different users), and are also possibly too rigid. Software descriptors, on the other hand, are arguably too general, and also difficult to find in the code because they are used to mean various things.

In the end, we decided on a "specific" software descriptor, findex, roughly standing for "Floris index" but possibly pronounced "findex" as it reads. This allows:

• specificity over more general software terms such as row, index, etc.
• flexibility as to the precise physical definition of what appears in the zeroth dimension
• ease of finding in the code, since findex is not used elsewhere in the code.

Additionally, the length of the findex dimension will be denoted n_findex.

2. Description of dimension 0

We'll also need a way to describe the zeroth dimension in documentation and in-code comments at the top level (e.g. examples, possibly FlorisInterface). We have decided, for now, to leave this open and see how we end up referring to the dimension as the development of the low-level code progresses, because we anticipate that the documentation and examples will be the last components to be finalized for v4.0. This will very likely be a physical descriptor, such as "sample", "condition", "state", or "event".

Please add comments to provide other options in the discussion.

[misi9170]

See also the pull request rafmudaf#60 but please continue the discussion here rather than in that pull request.