Linux on Arm fails with '-headersclean' in global/qfloat16.h

  /home/qt/work/qt/qtbase/src/corelib/global/qfloat16.h: In function
  ‘qfloat16::NativeType operator+(qfloat16, qfloat16::NativeType)’:
  /home/qt/work/qt/qtbase/src/corelib/global/qfloat16.h:178:31: error:
  conversion from ‘float’ to ‘qfloat16::NativeType’ {aka
  ‘__fp16’} may change value [-Werror=float-conversion]
    178 |     QF16_MAKE_ARITH_OP_FP(FP, +) \
        |                               ^
  /home/qt/work/qt/qtbase/src/corelib/global/qfloat16.h:172:95: note: in
  definition of macro ‘QF16_MAKE_ARITH_OP_FP’
    172 |     friend inline FP operator OP(qfloat16 lhs, FP rhs) noexcept { return static_cast<FP>(lhs) OP rhs; } \
        |                                                                                               ^~
  /home/qt/work/qt/qtbase/src/corelib/global/qfloat16.h:191:5: note: in
  expansion of macro ‘QF16_MAKE_ARITH_OP’
    191 |     QF16_MAKE_ARITH_OP(NativeType)
        |     ^~~~~~~~~~~~~~~~~~
  /home/qt/work/qt/qtbase/src/corelib/global/qfloat16.h: In function
  ‘qfloat16::NativeType operator+(qfloat16::NativeType, qfloat16)’:
  /home/qt/work/qt/qtbase/src/corelib/global/qfloat16.h:178:31: error:
  conversion from ‘float’ to ‘qfloat16::NativeType’ {aka
  ‘__fp16’} may change value [-Werror=float-conversion]
    178 |     QF16_MAKE_ARITH_OP_FP(FP, +) \
    ...