No, they are not from headers ( those do not have the golden inner "springs" ) but from a pcb out of some medical instrument. But you can use the pins from a good IC socket too.
Every Vref is usefull to calibrate something as long as it's value is much better known as that somethings accuracy.
You can build something with 4 TL431 like I did, it makes 10V but without reference or calibration you will never know the value.
A reference itself had several things. It has a typical value. Most can en will deviade a few mV from that. They most times come in several grades. The more change it is close to typical, the more expensive. The LT1027 for instance is in a A,B and C grade. The A grade is very hard to find, most sell the C grade, LT only samples the B, the A is the best.
So that makes a TL431 ( or other simple reference) without calibration only usefull for a cheap 2,5 digit multimeter or AVO 8 or so.
A reference needs time to stabilize. The fist few hunderd hours they change the most
The change by temp, relative humidity etc, but that is in the tens/hundereds of uV region. Not important for a 3,5 digit 1% meter.
Most references do not like o be loaded direct by a meter, they need a buffer amp behind them.
I have 7,5 digit meters, calibrators and standards so I can adjust and monitor a Vref with some confidence. Problem is the climate change in my lab. This gives me a 100 uV uncertanty. I have a climate independed knowh source so with a lot of work and enough speed I can bring it down to less as 5 uV