Refactor inlining logic into try_inline function

This commit extracts the logic for inlining function calls into a new
private
method, `try_inline`. This improves code organization and reduces
duplication in the `Optimizer`'s `visit_node` method.

The extracted method handles the preparation of substitutions and the
recursive visiting of the inlined node. It also ensures that the
substitution map state is correctly synchronized back to the calling
context.

examples/sma.myc

@@ -5,6 +5,8 @@
 
   (macro printx [s] `(fn [x] (print ~s x)))
 
-  (pipe [(.close src)] (printx "close=")) 
-  (pipe [(pipe [(.close src)] (SMA 20))] (printx "sma="))
+  (pipe src print) 
+  (pipe [(pipe [(.close src)] (SMA 20))] (printx "sma20="))
+  (pipe [(pipe [(.close src)] (SMA 5))] (printx "sma5="))
+  (pipe [(pipe [(.close src)] (SMA 100))] (printx "sma100="))
 )