Pica: Set program code / swizzle data limit to 4096

One of the later commits will enable writing to GS regs.
It turns out that on startup, most games will write 4096 GS program words.

The current limit of 1024 would hence result in 3072 (4096 - 1024) error messages:
```
HW.GPU <Error> video_core/shader/shader.cpp:WriteProgramCode:229: Invalid GS program offset 1024
```

New constants have been introduced to represent these limits.
The swizzle data size has also been raised. This matches the given field sizes of [GPUREG_SH_OPDESCS_INDEX](https://3dbrew.org/wiki/GPU/Internal_Registers#GPUREG_SH_OPDESCS_INDEX) and [GPUREG_SH_CODETRANSFER_INDEX](https://www.3dbrew.org/wiki/GPU/Internal_Registers#GPUREG_SH_CODETRANSFER_INDEX) (12 bit = [0; 4095]).

src/video_core/shader/shader.h

@@ -24,6 +24,9 @@ namespace Pica {
 
 namespace Shader {
 
+constexpr unsigned MAX_PROGRAM_CODE_LENGTH = 4096;
+constexpr unsigned MAX_SWIZZLE_DATA_LENGTH = 4096;
+
 struct AttributeBuffer {
     alignas(16) Math::Vec4<float24> attr[16];
 };
@@ -144,8 +147,8 @@ struct ShaderSetup {
         return offsetof(ShaderSetup, uniforms.i) + index * sizeof(Math::Vec4<u8>);
     }
 
-    std::array<u32, 1024> program_code;
+    std::array<u32, MAX_PROGRAM_CODE_LENGTH> program_code;
-    std::array<u32, 1024> swizzle_data;
+    std::array<u32, MAX_SWIZZLE_DATA_LENGTH> swizzle_data;
 
     /// Data private to ShaderEngines
     struct EngineData {

src/video_core/shader/shader_interpreter.cpp

@@ -653,7 +653,7 @@ static void RunInterpreter(const ShaderSetup& setup, UnitState& state, DebugData
 }
 
 void InterpreterEngine::SetupBatch(ShaderSetup& setup, unsigned int entry_point) {
-    ASSERT(entry_point < 1024);
+    ASSERT(entry_point < MAX_PROGRAM_CODE_LENGTH);
     setup.engine_data.entry_point = entry_point;
 }
 

src/video_core/shader/shader_jit_x64.cpp

@@ -15,7 +15,7 @@ JitX64Engine::JitX64Engine() = default;
 JitX64Engine::~JitX64Engine() = default;
 
 void JitX64Engine::SetupBatch(ShaderSetup& setup, unsigned int entry_point) {
-    ASSERT(entry_point < 1024);
+    ASSERT(entry_point < MAX_PROGRAM_CODE_LENGTH);
     setup.engine_data.entry_point = entry_point;
 
     u64 code_hash = Common::ComputeHash64(&setup.program_code, sizeof(setup.program_code));

src/video_core/shader/shader_jit_x64_compiler.cpp

@@ -834,8 +834,8 @@ void JitShader::FindReturnOffsets() {
     std::sort(return_offsets.begin(), return_offsets.end());
 }
 
-void JitShader::Compile(const std::array<u32, 1024>* program_code_,
+void JitShader::Compile(const std::array<u32, MAX_PROGRAM_CODE_LENGTH>* program_code_,
-                        const std::array<u32, 1024>* swizzle_data_) {
+                        const std::array<u32, MAX_SWIZZLE_DATA_LENGTH>* swizzle_data_) {
     program_code = program_code_;
     swizzle_data = swizzle_data_;
 

src/video_core/shader/shader_jit_x64_compiler.h

@@ -22,8 +22,8 @@ namespace Pica {
 
 namespace Shader {
 
-/// Memory allocated for each compiled shader (64Kb)
+/// Memory allocated for each compiled shader
-constexpr size_t MAX_SHADER_SIZE = 1024 * 64;
+constexpr size_t MAX_SHADER_SIZE = MAX_PROGRAM_CODE_LENGTH * 64;
 
 /**
  * This class implements the shader JIT compiler. It recompiles a Pica shader program into x86_64
@@ -37,8 +37,8 @@ public:
         program(&setup, &state, instruction_labels[offset].getAddress());
     }
 
-    void Compile(const std::array<u32, 1024>* program_code,
+    void Compile(const std::array<u32, MAX_PROGRAM_CODE_LENGTH>* program_code,
-                 const std::array<u32, 1024>* swizzle_data);
+                 const std::array<u32, MAX_SWIZZLE_DATA_LENGTH>* swizzle_data);
 
     void Compile_ADD(Instruction instr);
     void Compile_DP3(Instruction instr);
@@ -104,11 +104,11 @@ private:
      */
     void FindReturnOffsets();
 
-    const std::array<u32, 1024>* program_code = nullptr;
+    const std::array<u32, MAX_PROGRAM_CODE_LENGTH>* program_code = nullptr;
-    const std::array<u32, 1024>* swizzle_data = nullptr;
+    const std::array<u32, MAX_SWIZZLE_DATA_LENGTH>* swizzle_data = nullptr;
 
     /// Mapping of Pica VS instructions to pointers in the emitted code
-    std::array<Xbyak::Label, 1024> instruction_labels;
+    std::array<Xbyak::Label, MAX_PROGRAM_CODE_LENGTH> instruction_labels;
 
     /// Offsets in code where a return needs to be inserted
     std::vector<unsigned> return_offsets;