If your editor knows about comments, then the simplest method is to put a comment character on the blank lines:
lorem ipsums
%
\begin{align}
foo & bar\\
qux & rofl\\
& toto
\end{align}
%
lorem ipsums
But this will only work if your editor knows not to break up comment lines.
Update: Although I think that the real answer is: use a formatter that's LaTeX-aware, here's three answers to the actual question. This may not be very elegant, and it may break something elsewhere (I haven't tested it much).
\documentclass{article}
% \url{http://tex.stackexchange.com/q/24786/86}
\let\savepar=\par
\def\restorepar{\let\par=\savepar}
\def\nopar{\let\par=\restorepar}
\long\def\igpar#1{}
% Equivalent to the above is:
%\makeatletter
%\let\igpar=\@gobble
%\makeatother
\makeatletter
\def\eatpar{\@ifnextchar\par{\@gobble}{}}
\makeatother
\begin{document}
This is a paragraph.
This is the next paragraph.
This should be part of the previous one.
This is a paragraph.
This is the next paragraph.
\nopar
This should be part of the previous one.
This is a paragraph.
This is the next paragraph.
\igpar
This should be part of the previous one.
This is a paragraph.
This is the next paragraph.
\eatpar
This should be part of the previous one.
\end{document}
I recommend that these be on their own on a line. Not only does that make it easier to remove them later, but also means that the spacing at the end of the previous line is preserved.
The first (\nopar) works by redefining \par so that the next occurrence of \par is effectively ignored. So that it only works once, the next occurrence of \par also resets itself back to \par so that the next one will work as before. The \nopar can actually occur anywhere, not just before the blank line.
The second (\igpar) simply gobbles its next argument. As it is \long, it can gobble a \par with impunity.
The third (\eatpar) tests the next token to see if it is a \par or not. If it is, it gobbles it. If not, it leaves it alone. This is probably the most robust.
All of them work because of how TeX treats blank lines. When the input processor of TeX sees a blank line, it inserts a \par token in to the stream. This means that the blank line gets replaced by a \par before any macro processing is done. So in the internals of TeX, we can test for blank lines by testing against \pars. Otherwise, we'd have to test for the newline character, and test to see if we got two in a row.
