rework ShaderCompilerService to improve performance

- remove support for non-shared contextes parallel compilation.
  this wasn't used. we can always revive it later if we need to.

- rework how callbacks work so that we don't have to use a work list
  executed at each tick() in the shared context case (common case).
  this improves performance significantly on low-end devices, by
  not having to go through the list to check if all programs are
  compiled, multiple times per frame.

The new CallbackManager handles scheduling the callbacks after all
previous programs are compiled.

filament/backend/src/CompilerThreadPool.cpp

@@ -16,6 +16,8 @@
 
 #include "CompilerThreadPool.h"
 
+#include <utils/Systrace.h>
+
 #include <memory>
 
 namespace filament::backend {
@@ -38,6 +40,7 @@ void CompilerThreadPool::init(uint32_t threadCount, JobSystem::Priority priority
 
     for (size_t i = 0; i < threadCount; i++) {
         mCompilerThreads.emplace_back([this, priority, setup]() {
+            SYSTRACE_CONTEXT();
             // give the thread a name
             JobSystem::setThreadName("CompilerThreadPool");
             // run at a slightly lower priority than other filament threads
@@ -53,7 +56,11 @@ void CompilerThreadPool::init(uint32_t threadCount, JobSystem::Priority priority
                             (!std::all_of( std::begin(mQueues), std::end(mQueues),
                                     [](auto&& q) { return q.empty(); }));
                 });
-                if (!mExitRequested) {
+
+                SYSTRACE_VALUE32("CompilerThreadPool Jobs",
+                        mQueues[0].size() + mQueues[1].size());
+
+                if (UTILS_LIKELY(!mExitRequested)) {
                     Job job;
                     // use the first queue that's not empty
                     auto& queue = [this]() -> auto& {