[sargent_surico] Update arviz calls for the 1.x API

The lecture installs arviz unpinned, and `pip install arviz` now resolves
1.2.0 -- the rewrite that dispatches to arviz-base/arviz-stats/arviz-plots.
Two calls still used the 0.x API.

`az.from_dict` took the posterior group as its first positional argument in
0.x, so a flat {var_name: array} mapping worked. In 1.x the first argument is
{group_name: {var_name: array}}, so each entry of FREE was read as a group and
dict_to_dataset() was handed a raw ndarray, failing the HTML build with
`AttributeError: 'numpy.ndarray' object has no attribute 'items'`.

`az.summary` no longer emits hdi_3%/hdi_97%; it defaults to an 89% ETI. Ask
for ci_kind='hdi', ci_prob=0.94 to keep the interval the 0.x default gave, and
select the new hdi94_lb/hdi94_ub columns. Execution stopped at the from_dict
cell above, so this second break never reached the CI log.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

Author: mmcky

lectures/sargent_surico.md

@@ -1284,7 +1284,8 @@ A correlated chain of length $N$ is worth fewer than $N$ independent draws, and
 ```{code-cell} ipython3
 import arviz as az
 
-rwmh_idata = az.from_dict({n: kept[:, j][None, :] for j, n in enumerate(FREE)})
+rwmh_idata = az.from_dict(
+    {'posterior': {n: kept[:, j][None, :] for j, n in enumerate(FREE)}})
 rwmh_summary = az.summary(rwmh_idata, var_names=FREE)
 print(rwmh_summary[['mean', 'sd', 'ess_bulk']].to_string())
 print(f'\n{len(kept)} retained draws out of {N_DRAWS}, in {rwmh_seconds:.0f} seconds')
@@ -1637,9 +1638,10 @@ print(f'divergences                        '
 
 ```{code-cell} ipython3
 nuts_idata = az.from_numpyro(mcmc)
-nuts_summary = az.summary(nuts_idata, var_names=FREE)
+nuts_summary = az.summary(nuts_idata, var_names=FREE,
+                          ci_kind='hdi', ci_prob=0.94)
 nuts_kept = np.column_stack([np.asarray(mcmc.get_samples()[n]) for n in FREE])
-nuts_summary[['mean', 'sd', 'hdi_3%', 'hdi_97%', 'ess_bulk', 'r_hat']]
+nuts_summary[['mean', 'sd', 'hdi94_lb', 'hdi94_ub', 'ess_bulk', 'r_hat']]
 ```
 
 ### A warning sign in the diagnostics