You are entirely correct. Todd and Engi write:
The mineral assemblages of rocks at the surface lie on a locus of
P–T-conditions… It must be stressed…does not imply a P–T-distribution or geotherm realized at any one time during the orogenic evolution of this area. Even assuming the P–T-values
reflect maximum metamorphic conditions, these were established at
different times in different areas…show the integrated effect of all
the tectonometamorphic processes…i.e. the sum of burial, heating,
tectonic and erosional unroofing and cooling.
Notwithstanding, the metamorphic field gradient is still useful, particularly when comparing different sets of rocks.
Each of the samples in your figure (A, B and C) is a single rock and looking at it tells you the history of this rock and this rock alone. When studying a large area (a "field") there will be many rocks, that may differ from each other in their individual P–T-path. For example, in a rock suite that is now dipping 90°, one might argue that walking 3 kilometres on the surface equals 1 kbar of pressure, equals about 3 kilometres of depth. Because of that, each sample along the transect will have experienced different temperatures, with ideally the deeper ones also being the hotter ones.
By choosing a convenient reference point (maximum temperature of an individual sample), and connecting all samples along a transect, you can get an estimate of a P–T gradient, which can approximate a geotherm (under many assumptions). The same Todd and Engi paper has this figure:

Each of the lines is based on several dozens of individual samples. Based on this representation of the data, they go on to conclude:
The metamorphic field gradients and age determinations and consistent
with two stages of metamorphism. The first stage created the field
gradients preserved in the northern part of the field area. The second
thermal pulse affected only the south-eastern part of the field area.
This occurred at a later time, under lower pressure conditions.
Reaching those conclusions is only possible when condensing the 100 or so individual samples they used (i.e. 100 individual P–T-path loops) to what turned out to be only two metamorphic field gradients.
Todd, C. S. and Engi, M. (1997), Metamorphic field gradients in the Central Alps. Journal of Metamorphic Geology, 15: 513-530. doi:10.1111/j.1525-1314.1997.00038.x