Complete imports work by running sequentially data import for each imported project. This is a consequence of using the Jira "restore project from backup" functionality that is restricted to importing data for a single project.
An inconvenience derived from this is that, for each imported project, a number of processes must be run (mapping, validation, data import itself, indexing). This means that there is a minimum time required for each project, even if the volume of data in each project is not large.
This means, in a migration with a very large number of projects but not many issues overall, the performance would be poor.
With this improvement, indexing is now postponed until the end of the process, when data for all projects have been imported.