Implement datapar for parallel algorithms

[hkaiser]

Here is the list of all parallel algorithms. We plan to enable most (if not all) to support datapar, datapar(task) and variations.

• [x] adjacent_difference (#5580)
• [x] inner_product (see #2330)
• [x] adjacent_find (#5741 )
• [x] all_of any_of none_of (#5638)
• [x] copy copy_n (#5428)
• [ ] copy_if
• [ ] move
• [x] count count_if (see #2330)
• [x] equal mismatch (#5754, #5755)
• [ ] exclusive_scan inclusive_scan
• [x] fill fill_n (#5465)
• [x] find find_end find_first_of find_if find_if_not (#5740)
• [x] for_each for_each_n (see #2330)
• [x] generate generate_n (#5594)
• [ ] is_heap is_heap_until
• [ ] is_partitioned is_sorted is_sorted_until
• [ ] lexicographical_compare
• [ ] max_element min_element minmax_element
• [ ] partial_sort partial_sort_copy nth_element sort stable_sort
• [ ] partition partition_copy stable_partition
• [ ] remove remove_if
• [ ] remove_copy remove_copy_if
• [x] reduce
• [x] replace replace_copy replace_copy_if replace_if
• [ ] reverse reverse_copy
• [ ] rotate rotate_copy
• [ ] search search_n
• [ ] set_difference set_intersection set_symmetric_difference set_union includes
• [ ] inplace_merge merge
• [ ] swap_ranges
• [x] transform (see #2330)
• [x] transform_reduce
• [ ] transform_exclusive_scan transform_inclusive_scan
• [ ] uninitialized_copy uninitialized_copy_n
• [ ] uninitialized_fill uninitialized_fill_n
• [ ] unique unique_copy

[msimberg]

Removing the milestone as the individual issues can be scheduled for milestones instead (and it's unlikely that all of them will be done for one milestone).

[hkaiser]

This is not done yet - reopening.