[lmi] Ideal composite issue age

[Greg Chicares]

What (phony) issue age should a composite have?

Test case:
  File | New | Census
  Census | Run case

Right now the composite issue age is zero--so, on the calculation summary,
"Attained Age" starts at 0 and goes to 54. The starting "age" is zero, and
it's pretty easy to guess why. And the final "age", 54, only takes a little
thought to figure out, if we're using the default age of 45. However, if the
lowest age is 23, then the composite's last "Attained Age" is 76, and the
reason for that is not immediately obvious.

Instead, we'll deem the composite's issue age to be the lowest issue age in
any cell *that's included in the composite*. Then, in the test case above,
"Attained Age" would start at 45 and go to 99. And if the composite starts
at age 23, the explanation is easy to see: that's the lowest age.

(From my personal perspective: six of one, half a dozen of another; but I
have it on reliable authority that the new way will be clearer to end users.)

I'll make it so in svn soon. But first I'll commit a refinement to the
20151108T1257Z "ideal composite length". Motivating test case:

cd /opt/lmi/bin
./lmi_wx_shared --ash_nazg --data_path=/opt/lmi/data
  Alt-F N C
  Ctrl-+
  Ctrl-E
    "Issue age": 35
    OK
  Ctrl-+
  Ctrl-E
    "Issue age": 32
    uncheck "Include in composite"
    OK
  Alt-F A 'lowest_age_ignored.cns'
    Save
  Ctrl-Shift-R
    [looks like it's 65 years long...but is it?]
  Alt-Spacebar C

./lmi_cli_shared --accept --ash_nazg \
  --data_path=/opt/lmi/data \
  --file=/opt/lmi/eraseme/lowest_age_ignored.cns \
  --emit=emit_quietly,emit_test_data,emit_composite_only
less -N lowest_age_ignored.composite.000000000.test

Nope, it's 68 years long. We use 'emit_test_data' for regression testing,
so we don't want the three extra years.