Diffraction grating: Why does the light need to be in-phase?

I am currently studying the textbook Infrared and Raman Spectroscopy, 2nd edition, by Peter J. Larkin. Section 1.1 Dispersive Systems of chapter 3 says the following:

A monochromator consists of an entrance slit, followed by a mirror to insure the light is parallel, a diffraction grating, and a focusing mirror, which directs the dispersed radiation to the exit slit and onto a detector. In a scanning type monochromator, a scanning mechanism passes the dispersed radiation over a slit that isolates the frequency range falling on the detector. This type of instrument has limited sensitivity since at any one time, most of the light does not reach the detector.
Polychromatic radiation is sorted spatially into monochromatic components using a diffraction grating to bend the radiation by an angle that varies with wavelength. The diffraction grating contains many parallel lines (or grooves) on a reflective planar or concave support that are spaced a distance similar to the wavelength of light to be analysed. Incident radiation approaching the adjacent grooves in-phase is reflected with a plath length difference. The path length difference depends on the groove spacing, the angle of incidence ($alpha$), and the angle of reflectance of the radiation ($beta$).

Fig. 3.1 shows a schematic of a diffraction grating with the incident polychromatic radiation and the resultant diffracted light. When the in-phase incident radiation is reflected from the grating, radiation of suitable wavelength is focused onto the exit slit. At the exit slit the focused radiation will be in-phase for only a selected wavelength and its whole number multiples, which will constructively interfere and pass through the exit slit. Other wavelengths will destructively interfere and will not exit the monochromator. Thus, each of the grooves acts as an individual slitlike source of radiation, diffracting it in various directions. Typically, a selective filter is used to remove the higher-order wavelengths. When the grating is slightly rotated a slightly different wavelength will reach the detector.

I’m not sure that I’m completely understanding why the light needs to be in-phase. I understand that light being in-phase is a requirement for constructive interference, but why is this constructive interference then necessary to the process? I would greatly appreciate it if people would please take the time to explain this more clearly.