Support fp8 dynamic quantization for fmha (#3206)

* Support qscale for dynamic quant, remove static quant

* Support hdim=256

* Remove bias test case for fp8

---------

Co-authored-by: Illia Silin <98187287+illsilin@users.noreply.github.com>
Co-authored-by: asleepzzz <hanwen.chang@amd.com>

include/ck_tile/host/host_tensor.hpp

@@ -598,6 +598,8 @@ struct HostTensor
 
     typename Data::size_type size() const { return mData.size(); }
 
+    T max() const { return *std::max_element(mData.begin(), mData.end()); }
+
     // return a slice of this tensor
     // for simplicity we just copy the data and return a new tensor
     auto slice(std::vector<size_t> s_begin, std::vector<size_t> s_end) const