Move everything except public functionality to detail namespace

[msimberg]

The public API of pika is small: sender/receiver functionality, runtime initialization, what else?

Hidden functionality can then gradually be brought into the public namespace through pika::experimental:: or directly into pika::.

The only reasonable way to do this is module by module:

• [x] affinity (#152)
• [x] algorithms (#377, #411, #475)
• [x] allocator_support (#153)
• [x] assertion (#155)
• [x] async_base (#158)
• [x] async_combinators (#160)
• [x] async_cuda (#196)
• [x] async (nothing to be done)
• [x] async_mpi (#374)
• [x] command_line_handling (#216)
• [x] concepts (nothing to be done)
• [x] concurrency (#246)
• [x] config (#248)
• [x] coroutines (#257)
• [x] datastructures (#276)
• [x] debugging (#324)
• [x] errors (#365)
• [x] execution(#508)
• [x] execution_base (#508)
• [x] executors (#508)
• [x] filesystem (#379)
• [x] format (#487)
• [x] functional (#380)
• [x] futures (#525)
• [x] hardware (#607)
• [x] hashing (#631)
• [x] include (#632)
• [x] ini (#633)
• [x] init_runtime (#634)
• [ ] iterator_support
• [x] itt_notify (#1290)
• [ ] lock_registration
• [ ] logging
• [ ] memory
• [ ] mpi_base
• [ ] pack_traversal
• [ ] prefix (#1177)
• [ ] preprocessor
• [ ] program_options
• [x] properties (#673)
• [ ] resource_partitioner
• [ ] runtime_configuration
• [x] runtime (#826, #1091)
• [x] schedulers (#625)
• [x] string_util (#595)
• [x] synchronization (#483)
• [x] tag_invoke (#596) (#599)
• [x] testing (#594)
• [x] thread_pool_util (#509)
• [x] thread_pools (#462)
• [x] thread_support (#461)
• [x] threading (nothing to be done)
• [x] threading_base (#445)
• [x] threadmanager (#428)
• [x] timing (#209)
• [x] topology (#179)
• [x] type_support (#386, #400)
• [x] util (#420)
• [x] version (#166)

This is also a good opportunity to do general cleanup. Avoid nesting detail namespaces into experimental namespace #448