Fix PrMatchNN compilation on avx512 platforms

b4f8327c introduced a bug on platforms with AVX512 enabled because operator& is not implemented for the avx512 mask type. The solution is to use operator&& which is implemented for all platforms and also avoids the high-level notion of a bitwise AND. The ci-test does not catch this, because the platform is not tested by default. The build error can be seen in https://lhcb-nightlies.web.cern.ch/nightly/lhcb-head/3598/Rec/x86_64_v4+vecwid256-centos7-gcc11-opt/build.

Pr/PrAlgorithms/src/PrMatchNN.cpp

@@ -393,7 +393,7 @@ namespace LHCb::Pr::MatchNN {
       // If enabled filter on min track P, Pt
       if ( apply_min_p_cut ) {
         auto pMask = abs( qOverP ) < maxQOvP;
-        loopMask   = loopMask & pMask;
+        loopMask   = loopMask && pMask;
       }
       if ( apply_min_pt_cut ) {
         const auto momentum = abs( 1.0 / qOverP );
@@ -402,7 +402,7 @@ namespace LHCb::Pr::MatchNN {
         const auto sinTrack = sqrt( 1.0 - ( 1.0 / ( 1.0 + tx2 + ty2 ) ) );
         const auto pt       = sinTrack * momentum;
         auto       ptMask   = m_minPt.value() < pt;
-        loopMask            = loopMask & ptMask;
+        loopMask            = loopMask && ptMask;
       }
 
       auto const oTrack = result.compress_back<SIMDWrapper::InstructionSet::Best>( loopMask );