*: updated TODO

TODO

@@ -13,19 +13,13 @@ TODO:
 * review all NOLINT
 * bring nested groups back in place (see bd34e7f)
 * work stealing job system (see #100) + mt scheduler based on const awareness for types
-* view: reduce inst due to/improve perf with index-based approach in dispatch_get/pick_and_each/each (single type too, define storage ::at and ::at_as_tuple)
-* view: update natvis as needed after the last rework, merge pools/filter in the same array, drop check (?) and turn view into a position
-* view: type-only view_iterator (dyn get/excl sizes), type-only basic_common_view (dyn get/excl sizes with pointer to array from derived)
 * combine version-mask-vs-version-bits tricks with reserved bits to allow things like enabling/disabling
 * self contained entity traits to avoid explicit specializations (ie enum constants)
 * auto type info data from types if present
-* test: push sharing types further
 * storage entity: fast range-push from above
 * table: pop back to support swap and pop, single column access, empty type optimization
 * review cmake warning about FetchContent_Populate (need .28 and EXCLUDE_FROM_ALL for FetchContent)
 * suppress -Wself-move on CI with g++13
-* view specializations for multi, single and filtered elements
-* don't pass reactive storage by default to callback
 * runtime types support for meta for types that aren't backed by C++ types
 * built-in no-pagination storage - no_pagination page size as limits::max
 * any cdynamic to support const ownership construction
@@ -38,3 +32,4 @@ TODO:
 * avoid copying meta_type nodes, and use unique_ptr for meta_custom
 * paged vector as a standalone class
 * resource: shared_from_this?
+* meta perf: the node in the meta_any is pretty much irrelevant, but it affects perf a lot