#110 pdf.c has overall vertical offset issue

[Alan W. Irwin]

This overall (i.e., including all parts of the string including the part that is not superscripted or subscripted) vertical issue only occurs if "#u" (superscript command) appears anywhere in the string. It does not occur if "#d" (subscript command) occurs in the string without a "#u". So as a simple example of the issue (second page of test_superscript_subscript.py) if you generate the following three PLplot commands:

plmtex("t", -5.125, 0.1, 0.5, "X")
plmtex("t", -5.125, 0.2, 0.5, "X#d")
plmtex("t", -5.125, 0.3, 0.5, "X#u")

the third "X" is rendered lower than the other two. This result does not make any sense from the code perspective since there is no reason the #u should affect the position of the "X" behind it., and the #d and #u logic has essentially no differences except for the size and magnitude of the offset for _subsequent_ text. So possibly this puzzling result could be due to bad scaling (see previous pdf.c bug report).