It should be possible to prune historical state using a block height-based configuration setting. For example, anyone running your Data API implementation should be able to optionally configure the node to delete historical state related to blocks that are over 1000 blocks old. Storage pruning makes it easier for users with storage constraints to utilize your Data API implementation.
On a different note, large storage volumes can degrade RPC performance and lead to slower startup time (and can be generally harder to manage). For developers that extract all relevant information from the blockchain (and never access a block after it has been fetched once), limiting state storage can make for a more reliable integration.