Performance issue in chunkOnModifiedUtf8ByteSize

[rciovati]

While debugging some built time issues in our codebase I noticed that DeepLinkProcessor takes a suspiciously high amount of time (in our case 3x than Dagger). I grabbed a trace with YourKit and, if I'm reading it correctly, the processor is spending most of its time in chunkOnModifiedUtf8ByteSize.

We're not using ksp yet but it seems to me that this code path is shared between the apt and ksp so I suspect the issue exists there too.