HDRenderPipeline: Fix issue with specular color and packing precision

Assets/ScriptableRenderPipeline/HDRenderPipeline/Material/Lit/Lit.cs

         [GenerateHLSL]
         public enum SpecularValue
         {
+            // Value are defined in the tab name convertSpecularToValue in lit.hlsl
             Water = 0, // 0.02  Water or ice
             Regular = 1, // 0.04 regular dieletric
             Gemstone = 2, // 0.20

Assets/ScriptableRenderPipeline/HDRenderPipeline/Material/Lit/Lit.hlsl

 #endif
 }
 
-static const float3 arraySpecularValue = float3(SPECULARVALUE_WATER, SPECULARVALUE_REGULAR, SPECULARVALUE_GEMSTONE);
+static const float3 convertSpecularToValue = float3(0.02, 0.04, 0.20);
-    float val = arraySpecularValue[specular];
+    float val = convertSpecularToValue[specular];
     bsdfData.fresnel0 = lerp(val.xxx, baseColor, metallic);
 }
 
     {
         outGBuffer1.a = PackMaterialId(MATERIALID_LIT_STANDARD); // We save 1bit in gbuffer1 to store it in gbuffer2 instead
         // Encode specular on two bit for the enum, must match encoding of MATERIALID_LIT_STANDARD
-        // TODO: encoding here could be optimize as we know what is the value of surfaceData.specular
+        // TODO: encoding here could be optimize as we know what is the value of surfaceData.specular => (0.75294)
         outGBuffer2 = float4(surfaceData.specularColor, PackFloatInt8bit(0.0, surfaceData.specular, 4.0));
     }
     else if (surfaceData.materialId == MATERIALID_LIT_ANISO)
                 || anisotropy > 0)
         {
             bsdfData.materialId = MATERIALID_LIT_ANISO;
+            FillMaterialIdStandardData(baseColor, specular, metallic, bsdfData);
             float3 tangentWS = UnpackNormalOctEncode(float2(inGBuffer2.rg * 2.0 - 1.0));
             FillMaterialIdAnisoData(bsdfData.roughness, bsdfData.normalWS, tangentWS, anisotropy, bsdfData);
         }
         }
+
     }
     else // if (supportsSSS && bsdfData.materialId == MATERIALID_LIT_SSS)
     {

Assets/ScriptableRenderPipeline/ShaderLibrary/Packing.hlsl

     float t2 = (precision / maxi) / precisionMinusOne;
 
     // extract integer part
-    i = int(val / t2);
+    i = int((val / t2) + rcp(precisionMinusOne)); // + rcp(precisionMinusOne) to deal with precision issue (can't use round() as val contain the floating number
-    f = (-t2 * float(i) + val) / t1;
+    f = saturate((-t2 * float(i) + val) / t1); // Saturate in case of precision issue
-    return PackFloatInt(f, i, maxi, 255.0);
+    return PackFloatInt(f, i, maxi, 256.0);
-    UnpackFloatInt(val, maxi, 255.0, f, i);
+    UnpackFloatInt(val, maxi, 256.0, f, i);
 }
 
 float PackFloatInt10bit(float f, int i, float maxi)