To begin with, NIC interrupts are not just for packet reception. In example, think of link status change interrupts, error interrupts (when the network adapter needs to notify the host of a HW error) and other "housekeeping" ones. A complete list of mechanisms served by interrupts might be largely vendor-specific. My point is that one may not necessarily want to disable all interrupts; doing so may cause unwanted consequences.
With respect to Rx interrupts, it's interrupt mitigation which is commonly used. The idea is to let the NIC generate an Rx interrupt for a batch of packets rather than for a single received one (or, for larger batches compared to smaller ones). This way, the number of interrupts generated in a given time interval is reduced, which leads to lesser CPU time spent by the kernel in interrupt service routines (ISR).
On the other hand, disabling Rx interrupts is a completely different approach which implies that the kernel itself periodically polls the driver to check for new packets. In this scheme, it's not the network adapter which "awakens" the kernel by means of interrupts asking to process new packets; it's the kernel which on its own decides to check for new packets when it deems that necessary. This way, packet reception becomes even more efficient, and this is the main idea behind NAPI (New API).
What's for the specific driver which you're studying, one may not necessarily be well-versed in all the intricacies of it. It's better to file separate questions for specific device drivers.
