Fixes for the latest clang on osx (#14143)

Author: mpuccio

GPU/TPCFastTransformation/MultivariatePolynomialHelper.cxx

@@ -73,8 +73,7 @@ std::vector<std::string> MultivariatePolynomialHelper<0, 0, false>::getTerms() c
 TLinearFitter MultivariatePolynomialHelper<0, 0, false>::getTLinearFitter() const
 {
   const std::string formula = getTLinearFitterFormula();
-  TLinearFitter fitter(int32_t(mDim), formula.data(), "");
-  return fitter;
+  return TLinearFitter(int32_t(mDim), formula.data(), "");
 }
 
 std::vector<float> MultivariatePolynomialHelper<0, 0, false>::fit(TLinearFitter& fitter, std::vector<double>& x, std::vector<double>& y, std::vector<double>& error, const bool clearPoints)

Utilities/rANS/include/rANS/compat.h

@@ -201,7 +201,7 @@ class makeEncoder
   template <typename source_T>
   [[nodiscard]] inline static decltype(auto) fromSamples(gsl::span<const source_T> range, size_t renormingPrecision = 0)
   {
-    auto histogram = makeDenseHistogram::template fromSamples(range);
+    auto histogram = makeDenseHistogram::fromSamples(range);
     return makeEncoder::fromHistogram(std::move(histogram), renormingPrecision);
   };
 
@@ -296,4 +296,4 @@ using decoder_type = decltype(makeDecoder::fromRenormed(RenormedDenseHistogram<s
 
 } // namespace o2::rans::compat
 
-#endif /* RANS_COMPAT_H_ */
+#endif /* RANS_COMPAT_H_ */