Doesn't fail if id is not unique

[ckrf]

With a single id() variable that has duplicates, readreplace runs to completion instead of failing with an error.

[matthew-white]

This was a feature of readreplace version 1, which I maintained in version 2 for backwards compatibility reasons (see #22). readreplace completes all replacements in the replacements file in order, so if the id() variables aren't unique in the replacements file, the data set in memory will finish with the last replacement for each ID combination. The id() variables definitely aren't required to be unique in the data set in memory, as there are use cases outside data entry where users want to complete bulk replacements, completing the same replacement for multiple observations.

However, I could see this being a useful check for the replacements file. Maybe you could work with @boyercb on a pull request!