aco/live_var_analysis: use separate allocator for temporary live sets

This drastically reduces the memory footprint of the live sets.

Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29962>

src/amd/compiler/aco_live_var_analysis.cpp

@@ -68,6 +68,7 @@ get_temp_registers(aco_ptr<Instruction>& instr)
 namespace {
 
 struct live_ctx {
+   monotonic_buffer_resource m;
    Program* program;
    int worklist;
 };
@@ -92,7 +93,7 @@ process_live_temps_per_block(live_ctx& ctx, Block* block)
 {
    RegisterDemand new_demand;
    block->register_demand = RegisterDemand();
-   IDSet live = ctx.program->live.live_out[block->index];
+   IDSet live(ctx.program->live.live_out[block->index], ctx.m);
 
    /* initialize register demand */
    for (unsigned t : live)

src/amd/compiler/aco_util.h

@@ -514,6 +514,7 @@ struct IDSet {
    bool empty() const { return !size(); }
 
    explicit IDSet(monotonic_buffer_resource& m) : words(m) {}
+   explicit IDSet(const IDSet& other, monotonic_buffer_resource& m) : words(other.words, m) {}
 
 private:
    static uint32_t get_first_set(const block_t& words)