Use sparse multiport iterator

[edwardalee]

This makes use of the sparse iterator pull request 1298 of lingua-franca to get a performance improvement by using a sparse iterator over the input multiports.

[cmnrd]

Looks like we lost track of this PR. @edwardalee What should we do with this it? The @sparse annotation in Big does not compile anymore. Also, we have found a more general and more performant solution in the C++ target, that works on general multiports and does not require the @sparse annotation. The solution that Tassilo implemented in C++ (lf-lang/reactor-cpp#24) achieves more than 2x speedup in the Big benchmark. In comparison, the 15% improvement reported in lf-lang/lingua-franca#1298 seem marginal.

Should we drop this PR and also remove the @sparse annotation from LF. Or do you think it is worth fixing?

[edwardalee]

I've lost track of this also. The sparse access technique in the C target is documented and tested. See https://www.lf-lang.org/reactor-c/da/d00/port_8h.html and https://github.com/lf-lang/lingua-franca/blob/master/test/C/src/multiport/Sparse.lf and https://github.com/lf-lang/lingua-franca/blob/master/test/C/src/multiport/SparseFallback.lf

These do not use the @sparse annotation. Apparently, that is obsolete.