Why worry about the length of the pre-registration?
The premise behind pre-registrations is that interpreting results from statistical analyses requires knowing whether they are exploratory or confirmatory.

All readers of the results of an experiment need to also read the pre-registration to properly evaluate them. Not just the most expert readers, not just someone interested in replicating the study, not just the peer-review team. Everyone. Pre-registrations must hence be easy to find, and in a format that it is reasonable to expect all readers to consult and understand: short, standardized and precise.
This perspective on the role and appropriate nature of pre-registrations motivated the creation of AsPredicted.org.

Are character limits enforced?
The 3200 character limit is a recommendation, submissions with more characters are allowed. The system does enforce a limit based on the number of pages the final .pdf document would have (the link between number of characters and pages is imperfect due to various factors).

In particular:
If the pre-registration would be 1 page long, it can be submitted as is.
If the pre-registration would be 2 pages long, you will get a warning message and be asked to consider shortening it, but will be allowed to proceed.
If the pre-registration would be 3 pages long, you will not be allowed to submit it.

What to exclude from a pre-registration?
Details necessary to replicate a study or to examine the validity of the analyses or inferences are important and valuable, and should be included in the main manuscript, supplementary materials or posted as documentation alongside raw data and/or original materials, but in general they do not belong in a pre-registration.

When deciding whether to include information, this heuristic is useful:
Would a reader of the manuscript wonder whether a given decision about analysis, data source or hypothesis was made after knowing the results?
Only if the answer is yes, does information about it belong in the pre-registration.

For instance, imagine a study where participant rated attractiveness of people in various pictures. The pre-registration should indicate how many pictures were used, how attractiveness was measured, and which manipulations were used, but not which pictures were used, how they were chosen, the order they were shown in, if they were counterbalanced, how long they were shown for, size of the pictures, description of experimenter, whether there was a pilot study, run in groups or individually, etc. For most studies no reader would consider viable to selectively report any of these decisions based on results, and hence in most situations they do not belong in a pre-registration.