Atomic masses misinterpreted

Dear developers,

I happen to find that you may have some misinterpreted atomic masses raw data.

I know that the data is from [this repo](https://github.com/Bowserinator/Periodic-Table-JSON). Maybe I should also address an issue there. But let me speak here first, as we have [`Measurements.jl`](https://github.com/JuliaPhysics/Measurements.jl) by the same Organization.

[Atomic weights provided by IUPAC](https://www.ciaaw.org/atomic-weights.htm), which is the data source of Wikipedia, are given in two formats. The first one is `[x, y]`, means the measured value is between `x` and `y`, the second one is `d.dd(e)`, means the measured value is `d.dd`, with standard deviation of `0.0e`, or in short, `measurement("d.dd(e)")` if we use the `Measurements.jl` package.

However, `PeriodicTable.jl` gives
```julia
julia> elements[:He].atomic_mass
4.0026022 u
```
While the raw data from IUPAC is 4.002602(2). Apparently, when compile the data for `PeriodicTable.jl`, the error between parentheses is taken as the last decimal of the mean value, which means the value given by this package is always greater than that by IUPAC.

Hope we can fix it.

Cheers,
Yu

Start by tracing how the raw atomic-mass data is compiled into elements[:He].atomic_mass, comparing the reported 4.0026022 u with the IUPAC value 4.002602(2). Check the Measurements.jl representation and verify that the uncertainty is not appended to the mean value; done means the package output correctly reflects the source data.

Written by the indexing model from the issue text.