Don't guess time bounds for instantaneous fields

[ScottWales]

Describe the bug

When CSET loads a field it guesses bounds for all coordinates. This is not appropriate for instantaneous fields, and causes errors in the time values when loading the data with METplus where the times loaded from forecast data is in the middle of the correct instantaneous lead times.

How to reproduce

Steps to reproduce the behaviour:

Loading a file with air_temperature instantaneous, min, mean and max values with iris gives bounds only on the fields with cell_methods:

>>> iris_cubes = iris.load(path, "air_temperature")
>>> for c in iris_cubes:
...     print([str(m) for m in c.cell_methods], c.coord('time').bounds)
...
[] None
['time: minimum (interval: 1 hour)'] [[493032. 493033.]...[493041. 493042.]]
['time: mean (interval: 1 hour)'] [[493032. 493033.]...[493041. 493042.]]
['time: maximum (interval: 1 hour)'] [[493032. 493033.]...[493041. 493042.]]

Loading the same file with CSET gives bounds on everything, including the instantaneous field where they aren't appropriate:

>>> cset_cubes = CSET.operators.read.read_cubes(path, "air_temperature")
>>> for c in cset_cubes:
...     print([str(m) for m in c.cell_methods], c.coord('time').bounds)
...
[] [[493032.5 493033.5]...[493041.5 493042.5]]
['time: minimum (interval: 1 hour)'] [[493032. 493033.]...[493041. 493042.]]
['time: mean (interval: 1 hour)'] [[493032. 493033.]...[493041. 493042.]]
['time: maximum (interval: 1 hour)'] [[493032. 493033.]...[493041. 493042.]]

Expected behaviour

Fields without a time axis cell_method do not have bounds on the time axis