Like Health Information Exchange, FHIR, HL7 and the like, the great opportunity is to define semantic standards once and for all. Sure, data can be sent from EHR to EHR. All the data received is not useful if one does not really know or trust what it means. There is also the great opportunity for social determinants codification, the entire "omics" space, and patient generated information. All this will harken back to the age-old debates about a national patient identifier, central versus distributed data storage, vendor versus alternative options and the like. I would rather see the VA's dollars invested in moving this area forward.
Many EHR's have a lot of flexibility in both workflow and data collection, often times the amount of data documented is limited by the amount of effort IT/Executive Management is willing to invest in the process. Vendors are required to demonstrate advanced features for certification, hospitals are not required in all instances to implement them. For example, the complete picture of a patient would be a full list of immunizations, implantable devices, previous surgeries etc., not all facilities will invest the time to actually implement and thus complete the medical record.
I think that the technologies are getting pretty good with being able to include all sorts of data but health care systems, in particular, aren't great at sharing information with "competitors" (i.e., other care providers). Clinicians also aren't great at collecting social determinants of health, regardless of whether there's a place to capture the data in the EHR.
The last mile is always the toughest.
EMRs are just vessels. They are getting pretty good at incorporating information from other systems and presenting it to users. The problem is that our health care "system" is very messy and an EHR can't solve that. It can only present what it is made available to it, and it can only do so much to clean up the data without violating the original meaning intended by the original documenting provider.