MacOS build fixes

[srcejon]

A couple of fixes to M17 and DATV Demod so that it builds on MacOS Mohave.

[f4exb]

So if I get it correctly we will have soft LDPC decoding only for Linux (we drop it for Windows as well) ?

[srcejon]

No - I've just defined out some unused code as it doesn't compile on Mac. But yes, it is a bit confusing and could do with a further tidy up.

In datvdemodsink.cpp, before any of these LDPC changes, there was some code that was defined out:

#if 0
    // Doesn't work...
    // Soft LDPC decoder mode.
    // Deinterleave into soft bits.
    p_fecframes = new leansdr::pipebuf<leansdr::fecframe<leansdr::llr_sb> >(m_objScheduler, "FEC frames", BUF_FRAMES);
    p_s2_deinterleaver = new leansdr::s2_deinterleaver<leansdr::llr_ss, leansdr::llr_sb>(
        m_objScheduler,
        *(leansdr::pipebuf< leansdr::plslot<leansdr::llr_ss> > *) p_slots_dvbs2,
        *(leansdr::pipebuf< leansdr::fecframe<leansdr::llr_sb> > * ) p_fecframes
    );
    r_fecdecsoft = new leansdr::s2_fecdec_soft<leansdr::llr_t,leansdr::llr_sb>(
        m_objScheduler, *(leansdr::pipebuf< leansdr::fecframe<leansdr::llr_sb> > * ) p_fecframes,
        *(leansdr::pipebuf<leansdr::bbframe> *) p_bbframes,
        m_modcodModulation < 0 ? 0 : m_modcodModulation,
        true, 5,
        p_vbitcount,
        p_verrcount
    );
#else
    // External LDPC decoder mode.
    // Deinterleave into soft bits.
   ....

The bit of code I added the #ifdef LINUX for in this commit is only used inside the #if 0 here - so both bits of code could potentially be removed completely, unless you see a future need for them.

The "External LDPC decoder mode" above is the code that is currently used, even though the decoder is no longer external, so that could do with a bit of a tidy up as well.

We could also remove the USE_LDPC_TOOL #defined code as well. Depends if you just want to move forward with only the one implementation.

[f4exb]

OK I probably skipped some details indeed the original code is not very readable.

We could also remove the USE_LDPC_TOOL #defined code as well. Depends if you just want to move forward with only the one implementation.

In fact I have been testing the "internal" implementation with QO-100 DATV and it works at least as well as with the "external" implementation. There might even be some gain in processing since there is not the overhead of spawning a full process. So I think that eventually yes the old code could be removed.