[disclaimer: not a PacBio person, I'm just vaguely aware of the technology]
Movie files are the PacBio raw data format, representing observed signal intensities in each sequencing well over the course of the sequencing run. The movie files can be used to re-call sequences with updated statistical models, which can improve accuracy, or test out hypotheses about how epigenetics alters the rate of nucleotide incorporation (i.e. methylation).
Illumina has a similar static version of this which is an image (or multiple images, depending on chemistry) of the flow cell after each synthesis cycle. Illumina's raw data format consumes tens of terabytes per run, and is of minimal value after base calling, so is almost always discarded after base calling.
Anything uploaded to NCBI SRA must contain FASTQ information. While the original uploaded files can be kept (which may include time/signal information), the files are converted into a FASTQ-like format for NCBI's internal representation. As such, there's a chance that movie files could be uploaded to SRA, but those files would also need to include the called FASTQ information (e.g. in a HDF5 container). In contrast to this, ENA does allow uploading raw data files without called information.