Fix formatting and missing references in JOSS manuscript

paper.bib

@@ -48,3 +48,21 @@ @article{Clough:2005a
 	Volume = {91},
 	Year = {2005},
 }
+
+@article{Emanuel:1991a,
+	Author = {Kerry A. Emanuel},
+	Journal = {J. Atmos. Sci.},
+	Pages = {2313--2314},
+	Title = {A Scheme for Representing Cumulus Convection in Large-Scale Models},
+	Volume = {48},
+	Year = {1991},
+}
+
+@article{Emanuel:1999a,
+	Author = {Kerry A. Emanuel and Marina {\v Z}ivkovi{\'c}-Rothman},
+	Journal = {J. Atmos. Sci.},
+	Pages = {1766--1782},
+	Title = {Development and Evaluation of a Convection Scheme for Use in Climate Models},
+	Volume = {56},
+	Year = {1999},
+}

paper.md

@@ -19,13 +19,15 @@ bibliography: paper.bib
 CLIMLAB is an open-ended engine for interactive, process-oriented climate modeling for use in education and research. It is motivated by the need for simpler tools and more reproducible workflows with which to "fill in the gaps" between blackboard-level theory and the results of comprehensive climate models. With CLIMLAB you can interactively mix and match physical model components, or combine simpler process models together into a more comprehensive model. CLIMLAB is used in the classroom (undergraduate and graduate) to put models in the hands of students, and emphasize a hierarchical, process-oriented approach to understanding the key emergent properties of the climate system. CLIMLAB is equally a tool for climate research, where the same needs exist for more robust, process- based understanding and reproducible computational results [@Held:2005fk; @Jeevanjee:2017a].
 
 CLIMLAB defines a base Python class called `Process`. This generalized model operator contains, at a minimum:
+
 - a dictionary of state variables, each with a well-defined spatial domain
 - a list of required input fields
 - methods to compute tendencies (rates of change) of its state variables given inputs and current state.
 
 A `Process` object can also contain an arbitrarily complex tree of subprocesses (each also some sub-class of `Process`). Tendencies are then computed by iterating through the subprocess tree and summing up contributions from each member. Using this object-oriented approach, every climate process (radiative, dynamical, physical, turbulent, convective, chemical, etc.) can be simulated as a stand-alone model given appropriate input, or as a component of a more complex model.
 
-CLIMLAB has out-of-the-box support and documented examples for
+CLIMLAB has out-of-the-box support and documented examples for:
+
 - Several atmospheric radiation codes including the widely used [RRTMG model](http://rtweb.aer.com/rrtm_frame.html) [@Mlawer:1997a;@Clough:2005a]
 - Convection models including the [Emanuel moist convection scheme](https://emanuel.mit.edu/problem-convective-moistening) [@EmanueL:1991a; @EmanueL:1999a]
 - Diffusion solvers for energy balance models