Several things:
- Do you really really need to make 2.1 V to run the micro from the 3 V coin cell? There are many micros that will run nicely from the full range of voltage a "3V" coin cell will produce over its lifetime. A converter seems wasteful and unnecessary.
- You really expect a "coin cell" to drive a relay!? Unless these are very special tiny extra small low power minuature relays, that's not gonna happen. As soon as you attempt to activate a relay, the coin cell voltage collapses, which probably resets the CPU.
Since you say your relays are latching, you only need a pulse of power for a short time. One possibility is to put a large capacitor across the coin cell. This would hold up the voltage long enough for a relay to latch in the opposite state, which is 10 ms according to the datasheet you provided. However, that will still cause significant current drain on the battery to recharge the cap after the relay is tripped. That will decrease the lifetime of the coin cell. They are intended for µA continuous drain, and a few mA pulsed at most.
A better option for the battery is to charge up a capacitor to a higher voltage ahead of time, then discharge it across the relay. You can then control the switching power suply that charges the cap to only drain power from the coin cell slowly.
However, no matter what you do, basic physics limits how many times a single coin cell can activate the relay. It takes a finite amount of energy to flip the relay, and there is a finite (and small) amount of energy in a coin cell. Even with theoretical 100% conversion efficiency, that coin cell won't live long.
