[0pt,6pt] Clean up the prototype code for presenting to wider GR audience

[ivan-cukic]

Background

We need two types of processing: a) synchronous stream- as well as b) asynchronous message-type processing (i.e. chunked and/or DataSet based data). The original OpenDigitizer draft design intended that the latter would be done using RxCpp. However, since the RxCpp development somewhat stalled and because both processing types are graph-type problems, the proposal was to rationalise and merge both designs into one using GNU Radio in it's 4.0 version incarnation.

The graph-prototype is targeted as an enhancement and replacement of the low-level GNU Radio 4.0 API to enable such message-type processing using graphs.

Immediate things to be refactored prior to merging with GR 4.0 -- 'Step one'

• new/extended fluent connect API -> see Wrapping up the prototype for the new GR4 graph gnuradio4#25
• break graph connection into to steps -> see Wrapping up the prototype for the new GR4 graph gnuradio4#25
  • 'connect' (register new nodes and connections)
  • 'init' (establish buffer between nodes) -> buffer size -> either user-defined/default function that computes the appropriate size based on the ports' {MIN, MAX} input/output constraints
• check back with gr maintainers regarding vetting/use of refl-cpp -> @mormj & @RalphSteinhagen
• fluent design + ports as fields (idea by @dkozel) + settings reflection using refl-cpp, as a brief concept:

using namespace gr;
struct node_settings { // node domain object specialised/derived/modified (as needed)
    float samp_rate = 44100.f;
    int scaling_factor = 1;
};
ENABLE_REFLECTION_FOR(node_settings, samp_rate, scaling_factor) // <- refl::cpp macro

template<typename T>
class my_node : public node<my_node<T>>, Setting<node_settings> {
public:
    IN<T> in0;
    IN<T> in1;
    OUT<T> out;
    
    [[nodiscard]] constexpr T process_one(T value1, T value2) const noexcept {
        return  settings.scaling_factor * some_math_func(value1, value2); // ... do some operation with values from ports in[0,1]
    }
};
ENABLE_REFLECTION_FOR((my_node, float, double, int), in0, in1, out) // <- refl::cpp macro for my_node

• SIMD loop + epilogue integration - proof of concept : added self-contained SIMD tests to bm_case1 benchmarks gnuradio4#26
• pre-finish: @wirew0rm will iterate with @ivan-cukic and @RalphSteinhagen on the new refactored design
• finish: @wirew0rm (as lead) will iterate with GR maintainers (@mormj, @willcode et al.) regarding maintainability, readability, etc. (N.B. to be kept in mind... this is not feature complete yet, just the basic framework)
• post-finish: remove/close rxcpp issue w/ appropriate replacement comments
  (N.B. after GR governance team accepted to merge the ideas developed in graph-prototype with GR 4.0)
  • the following issues can be removed [total: 15SP]:
    [2pt] Investigate suitability of the RxCPP testing framework for use in OpenCMW opencmw-cpp#195
    [8pt] Implement RxCPP type-erased nodes and the dynamic pipeline loader opencmw-cpp#196
    [5pt] Implement compile-time meta-information for RxCPP pipelines opencmw-cpp#198
  • the following issues have to be adapted (total: 8SP):
    [3pt,0pt] Evaluate approaches for GR4-flowgraph integration with the OpenCMW Disruptor opencmw-cpp#199
    [5pt, 0pt] Implement gr4-flowgraph integration with the OpenCMW Disruptor opencmw-cpp#200

Steps afterward (EPIC: #42)

• immediate follow-up: [0pt] intermediate step: make graph-prototype pre OP-usable for FAIR/OpenDigitizer-UI #46

[ivan-cukic]

• the implementation is split into separate header files
• refactored the code
• implemented a nicer connection API
• split the work function into several reusable parts
• delay the connection logic until graph is explicitly initialized

[RalphSteinhagen]

Done. Thanks a lot to @ivan-cukic and the good work he put into this!