Fix bibstem computation for "spillover" volumes

[aaccomazzi]

We currently compute the bibstem for 2018SPIE10707E..09J as being SPIE1 rather than SPIE. Code that needs fixing is here: https://github.com/adsabs/ADSImportPipeline/blob/master/aip/classic/solr_adapter.py#L706

According to adspy, here is the list of affected publications:

# these are publications for which there may be a 5-digit volume
# which "spills left" i.e. has its most significant digit in the
# journal field
PUB_VOLUME_SPILLS_LEFT = (
    'SPIE',
    'ATel',
    'GCN',
)

[aaccomazzi]

Fixed via #225