Probably too late to comment since last post on this thread was a month ago.... but...
I got into the habit of using a series resistance between the collector and the coil. I pick a small resistance such that the voltage across the coil is still above the min spec holding voltage, considering it forms a voltage divider twixt the coil resistance and the resistor. Consider also the voltage is max when the transistor switches on (since i=0 initially), and reduces as current increases -- just what we want. Now the flyback diode is sometimes inside a driver (like ULN2003 or 2803) where I may tie the cathode side to 12V, for instance, while the coil is a 5V coil. I like to make use of such built in clamp diodes and the series resistor and option to clamp at higher voltage is what make a simple diode clamp work better (vis a vis the slowing of contact opening due to back emf).
The advantages are that the coil is not overdriven (keeps down coil temperature and overall current draw) which may be important. Still the initial voltage across the coil is high and falls to only as much as necessary to hold the contacts together. Then when the transistor switches off, the flyback current is not short circuited through the clamp diode so that the contacts can come apart quicker (so make sure your diode anode is on the other side of this resistor) not to mention also clamping to a higher voltage, allowing that much more time for the spring to pull the contacts apart. By and large, the contacts are more damaged by arcing while pulling apart than anything else and to avoid welded contacts may well save other components in a system. Seems worth it to apply more thought than the usual habit of throwing 1N4148s across coils, which is what I used to mindlessly do until I finally started looking at it some more.
My $0.02.