Unfortunately the situation is more complicated than dealing with the principal stresses (see this answer). The principal stress direction will only be $\pm 45^o$ when in a condition of pure shear. So for a beam this is at the neutral axis, away from the neutral axis the principal stress direction will change. This can be seen here:
Current concrete shear design takes a strut-and-tie approach. For this we assume a truss made from steel tension ties (reinforcement) and concrete compression struts:
In this image the dashed lines are in compression (which can be carried by the concrete), while the solid lines are in tension (which are carried by the reinforcement). According to Eurocode 2, the inclination of the compression structs is allowed to vary between $21.8^o$ and $45^o$, and depends on the shear present in the beam. This will vary along the length of the beam, resulting in the stirrup spacing changing.
Inclining the stirrups can achieve an increase in shear resistance. However, inclined stirrups are very unusual in practice. This is due to some of the disadvantages of inclined stirrups, for example:
- During load reversal (during an earthquake, for example), you would end up with effectively no shear reinforcement - which could result in failure.
- The extra expense/complication involved in placing inclined reinforcement is usually not worth the relatively minor increase in shear resistance.
- In most design situations the extra cost/time/risk involved in developing a non-standard design is discouraged except when absolutely necessary.