Spy-Bi-Wire vs JTAG

Spy-Bi-Wire is JTAG - the physical layer is just different. JTAG is a set of commands over a serial interface. The main problem is the chip you're debugging. The chip you're debugging has to have the support for the fancy features - not just your JTAG. And indeed, in TI's own documentation for the JTAG interface on MSP430 it spells out how its chips don't fully implement all JTAG commands (section 3.2, pg 67):

Only the BYPASS instruction is supported. There is no support for SAMPLE, PRELOAD, or EXTEST instructions.

Here's the fun part: since Spy-Bi-Wire is just serial communication, all of the actual cool features have to be implemented on the MSP430. For instance, the number of hardware breakpoints you have? Determined by the processor, not the JTAG you're using. Most MSP430's have two, some have eight.

I develop custom MSP430 solution that run on Teradyne ICT

My opinion, SbW is certainly not faster. It takes 3 clock cycles to accomplish 1 bit by JTAG. Then, somebody had a really bad idea to have the RESET line carry the data. Then the the RESET usually has at least 1nF capacitance, which makes it even slower to charge and discharge.

The SbW is fine for very tiny flash, like the Value Series MSP430G, having only a few K of flash to program. But for the MSP430F5XX and 6XX, you better provide connectivity to the entire JTAG set and dedicate those pins for programming, otherwise you will regret using the SbW.