fix: nested block on, make materialize async (#7618)

Calling `block_on` in `load_patches_to_gpu` caused a crash when running
the benchmark, as the bench runner also calls block on which led to
nested executors. Copies and async usage will be reworked holistically
as part of the bigger work stream making data loading fast.

Signed-off-by: Alexander Droste <alexander.droste@protonmail.com>

Author: 0ax1

vortex-cuda/benches/dynamic_dispatch_cuda.rs

@@ -17,6 +17,7 @@ use cudarc::driver::DevicePtr;
 use cudarc::driver::LaunchConfig;
 use cudarc::driver::PushKernelArg;
 use cudarc::driver::sys::CUevent_flags;
+use futures::executor::block_on;
 use vortex::array::IntoArray;
 use vortex::array::LEGACY_SESSION;
 use vortex::array::VortexSessionExecute;
@@ -133,7 +134,7 @@ impl BenchRunner {
             device_buffers,
             shared_mem_bytes,
             ..
-        } = plan.materialize(cuda_ctx).vortex_expect("materialize plan");
+        } = block_on(plan.materialize(cuda_ctx)).vortex_expect("materialize plan");
 
         let device_plan = Arc::new(
             cuda_ctx