Interrupt within an interrupt with 16F PIC is more difficult to code generally because all interrupts vectors to address 0x4.
This then involves two actions: saving context and testing & jump to correct handling routine. If one tests for the correct interrupt first, one would risk changing bits in STATUS or W registers which causes disaster later. So saving context has to be done first.
While the proper return address is taken care of by the PIC hardware stack, user must implement some form of software stack to save the context of registers otherwise the saving context action by the second interrupt will mess up the saved value of the first interrupt.
In case of PIC, it is so much simpler just to allow one interrupt to finish before processing the second interrupt. Multiple interrupts is common and easy to code using a software stack along with PUSH and POP instructions in other types of uP/uC.