Remove Sphinx cross-references from notebook and docs

Replaced Sphinx-style {ref} cross-references with plain text in code/notebook.ipynb and code/notebook.md for improved compatibility outside Sphinx. Added missing section anchors in content/paper/moderation.md to preserve reference targets. Updated PDF files with corresponding changes.

Author: alanlujan91

code/notebook.ipynb

@@ -312,7 +312,7 @@
     "\n",
     "### Figure 1: The EGM Extrapolation Problem\n",
     "\n",
-    "The **precautionary saving gap** (optimist minus realist consumption) must be positive: income risk induces additional saving. As {ref}`demonstrated in the paper <benchmark-the-method-of-endogenous-gridpoints>` and shown in [](#fig:egm-extrapolation-problem) ({ref}`Figure 1 <fig:ExtrapProblem>` in the paper), standard EGM violates this constraint when extrapolating {cite:p}`carrollEGM`."
+    "The **precautionary saving gap** (optimist minus realist consumption) must be positive: income risk induces additional saving. As demonstrated in the paper and shown in [](#fig:egm-extrapolation-problem), standard EGM violates this constraint when extrapolating {cite:p}`carrollEGM`."
    ]
   },
   {
@@ -483,7 +483,7 @@
     "5. Interpolate $\\logitModRte(\\logmNrmEx)$ with derivatives\n",
     "6. Reconstruct: $\\cFuncReal = \\cFuncPes + \\modRte \\cdot (\\cFuncOpt - \\cFuncPes)$\n",
     "\n",
-    "This ensures bound compliance via asymptotically linear extrapolation, as {ref}`derived in the paper <the-consumption-function>`. HARK uses **cubic Hermite interpolation** {cite:p}`Fritsch1980,FritschButland1984` for accuracy; see {cite:p}`Santos2000,JuddMaliarMaliar2017` on function approximation and error bounding.\n",
+    "This ensures bound compliance via asymptotically linear extrapolation, as derived in the paper. HARK uses **cubic Hermite interpolation** {cite:p}`Fritsch1980,FritschButland1984` for accuracy; see {cite:p}`Santos2000,JuddMaliarMaliar2017` on function approximation and error bounding.\n",
     "```\n",
     "\n",
     "```{note} The Transformation\n",

code/notebook.md

@@ -128,7 +128,7 @@ The first set of figures will focus on the core of the consumption-saving proble
 
 ### Figure 1: The EGM Extrapolation Problem
 
-The **precautionary saving gap** (optimist minus realist consumption) must be positive: income risk induces additional saving. As {ref}`demonstrated in the paper <benchmark-the-method-of-endogenous-gridpoints>` and shown in [](#fig:egm-extrapolation-problem) ({ref}`Figure 1 <fig:ExtrapProblem>` in the paper), standard EGM violates this constraint when extrapolating {cite:p}`carrollEGM`.
+The **precautionary saving gap** (optimist minus realist consumption) must be positive: income risk induces additional saving. As demonstrated in the paper and shown in [](#fig:egm-extrapolation-problem), standard EGM violates this constraint when extrapolating {cite:p}`carrollEGM`.
 
 ```python
 # | label: fig:egm-extrapolation-problem
@@ -209,7 +209,7 @@ MoM steps (notation matches the paper):
 5. Interpolate $\logitModRte(\logmNrmEx)$ with derivatives
 6. Reconstruct: $\cFuncReal = \cFuncPes + \modRte \cdot (\cFuncOpt - \cFuncPes)$
 
-This ensures bound compliance via asymptotically linear extrapolation, as {ref}`derived in the paper <the-consumption-function>`. HARK uses **cubic Hermite interpolation** {cite:p}`Fritsch1980,FritschButland1984` for accuracy; see {cite:p}`Santos2000,JuddMaliarMaliar2017` on function approximation and error bounding.
+This ensures bound compliance via asymptotically linear extrapolation, as derived in the paper. HARK uses **cubic Hermite interpolation** {cite:p}`Fritsch1980,FritschButland1984` for accuracy; see {cite:p}`Santos2000,JuddMaliarMaliar2017` on function approximation and error bounding.
 ```
 
 ```{note} The Transformation

content/paper/appendix_letters.pdf

@@ -143,6 +143,7 @@ For expositional simplicity in what follows, we set $\PermGroFac=1$ and assume $
 [^crra-not-one]:
     For notational simplicity, we henceforth assume $\CRRA \neq 1$. Most subsequent derivations involving transformations of the value function (such as $\vInvOpt$ and $\vInvReal$) contain expressions with denominators $(1-\CRRA)$ that are undefined when $\CRRA=1$. The case $\CRRA=1$ (logarithmic utility) requires parallel derivations that exploit the simplifications arising from $\log$ utility; the economic insights remain analogous.
 
+(benchmark-the-method-of-endogenous-gridpoints)=
 # Benchmark: The Method of Endogenous Gridpoints
 
 For comparison to our new solution method, we use the endogenous gridpoints
@@ -281,6 +282,7 @@ future income, and therefore consumes
 \end{aligned}
 ```
 
+(the-consumption-function)=
 ## The Consumption Function
 
 It seems obvious that the spending of the realist will be strictly greater than