doc: Spellcheck documentation ahead of 4.4 release. * aspell-octave.en.pws: Add new words to Octave-only spelling dictionary. Remove exception words which are no longer used in manual. * basics.txi, bugs.txi, func.txi, geometry.txi, install.txi, matrix.txi, package.txi, plot.txi, poly.txi, preface.txi, quad.txi, sparse.txi, strings.txi, vectorize.txi, data.cc, defaults.cc, file-io.cc, pinv.cc, quadcc.cc, qz.cc, rand.cc, schur.cc, syscalls.cc, sysdep.cc, toplev.cc, amd.cc, audioread.cc, colamd.cc, dmperm.cc, symrcm.cc, quadgk.m, quadl.m, imfinfo.m, rgb2gray.m, javachk.m, usejava.m, unpack.m, fzero.m, glpk.m, pqpnonneg.m, stemleaf.m, print.m, polyfit.m, blackman.m, bicgstab.m, cgs.m, eigs.m, pcg.m, tfqmr.m, gallery.m, rosser.m, toeplitz.m, vander.m, isstrprop.m: Add @nospell{} macro around proper names and other words which aspell should not check. Correct misspellings identified by aspell.

--- a/scripts/sparse/bicgstab.m	Mon Mar 26 22:54:24 2018 +0200
+++ b/scripts/sparse/bicgstab.m	Mon Mar 26 10:45:04 2018 -0700
@@ -53,7 +53,7 @@
 ## handle or inline function @code{g} such that
 ## @code{g(@var{x}) = @var{M1} \ @var{x}} or
 ## @code{g(@var{x}) = @var{M2} \ @var{x}}.
-## The techinque used is the right preconditioning, i.e., it is
+## The technique used is the right preconditioning, i.e., it is
 ## solved @code{@var{A} * inv (@var{M}) * @var{y} = @var{b}} and then
 ## @code{@var{x} = inv (@var{M}) * @var{y}}.
 ##
@@ -185,7 +185,7 @@
 ## ## reference solution computed by bicgstab after one iteration
 ## [x_ref, fl] = bicgstab (A, b, [], 1, M)
 ##
-## ## rigth preconditioning
+## ## right preconditioning
 ## [y, fl] = bicgstab (A / M, b, [], 1)
 ## x = M \ y # compare x and x_ref
 ##