Another issue with TRIACs is they have a much higher on loss, compared to relays which lose negligible power.
Relays can also fail short circuit, but it's less common than TRIACs which nearly always fail short circuit.
As far as reliability is concerned, there are some advantages TRIACs have over relays: they're less susceptible to vibration and the number of switching cycles is unlimited. In this case vibration can be mitigated by appropriate component selection and good mechanical design and the number switching cycles is a non-issue, as the motor won't be reversed frequently enough for that to matter.
In the case of controlling both the speed and direction of a universal motor is concerned: use a TRIAC for the former and relay for the latter. Using four TRIACs to control both the direction and power makes it more difficult to control, results in double the power dissipation, compared to a TIRAC and relay and if one of them fails short circuit, the power supply will be short circuited when the other one turns on. In short, using TRIACs to control the direction of a universal motor is a bad idea. Different components have their strengths and weaknesses and good engineering is about knowing which component is best suited to a certain application.
A latching relay for directional control is probably best, since there's no chance the motor can suddenly reverse, if the power to the relay coil fails. Reversing the motor when the shaft is moving at full speed, will result in an enormous peak current, nearly double the initial starting current which will cause total destruction of the driver circuitary.