Add optional convert options for r->clj functions

[behrica]

as discussed here:
https://clojurians.zulipchat.com/#narrow/stream/224816-clojisr-dev/topic/native.20support.20.20for.20tech.2Ev3.2Edataype.20tensors.20.3F/near/438179471

[behrica]

I did not change this function to take options.
Could not find any usage, neither sucessfully call it myself.

[genmeblog]

Why r should know about options?

[behrica]

hmm. Not sure any more. Was some time ago.
I think the idea was to "be able" to pass options through all layers.

[genmeblog]

Options should be passed only to the java->clj function.

Multidimensional conversion is done here: https://github.com/scicloj/clojisr/blob/master/src/clojisr/v1/impl/java_to_clj.clj#L94-L100

[behrica]

I don't remember any more about the details of the code changes.
The "goal" is still to be able later to implement this:

(require-r-package '[datasets])
(r->clj r.datasets/iris {:as-tensor true})

so we can offer options to change the build-in conversions.

[behrica]

@genmeblog
I have a first versin of a working conversion to tensor.
See here for test:

clojisr/src/clojisr/v1/r.clj

Line 242 in 79866c2

( r->clj {:as-tensor true}))

Maybe have a quick look at the flow of the "options"

[behrica]

I have not checked if the "order" of elements is correct.

[behrica]

I have the impression that this ns
tech.v3.tensor.dimensions

contains the right functions to "convert" in one go from the "matrix data as seq" as we get it from
(prot/->clj exp)
into the tensor....

[behrica]

@genmeblog
I noticed that for R arrays we alreday get a flat vector from REngine.
Is it somehwere documented whats the order of this is in case of multidimensional arrays ?

[behrica]

For reference:
https://cran.r-project.org/web/packages/reticulate/vignettes/arrays.html