Towards making smart-importer and beancount-import more cooperative.

beancount-import will now look for new_account and unconfirmed_account
meta fields in each posting, and understand their presence to mean that
the account names of these postings are tentative, potentially new.

Introduced two new methods similar to those present in training.py
and new logic.

beancount_import/reconcile.py

@@ -25,6 +25,8 @@
 
 from .matching import FIXME_ACCOUNT, is_unknown_account, CLEARED_KEY
 
+UNCONFIRMED_ACCOUNT_KEY = 'unconfirmed_account'
+NEW_ACCOUNT_KEY = 'new_account'
 display_prediction_explanation = False
 
 classifier_cache_version_number = 1
@@ -722,18 +724,53 @@ def _get_primary_transaction_amount_number(self, transaction: Transaction):
                 return -source_posting.units.number
         return None
 
+    def _get_unknown_account_names(self, transaction: Transaction):
+        return [posting.account for posting in transaction.postings if posting.meta is not None and NEW_ACCOUNT_KEY in posting.meta]
+
+    def _has_unconfirmed_account(self, transaction: Transaction) -> bool:
+        return any((posting.meta is not None and UNCONFIRMED_ACCOUNT_KEY in posting.meta)
+                for posting in transaction.postings)
+
+    def _group_predicted_accounts_by_name(self, transaction: Transaction):
+        '''
+            Takes a list of postings with candidate account names,
+            and groups them into groups that should share the same exact account.
+            Expects each predicted posting to have an UNCONFIRMED_ACCOUNT_KEY meta field.
+        '''
+        num_groups = 0
+        group_numbers = []
+        predicted_account_names = []
+        existing_groups = {}  # type: Dict[str, int]
+        new_accounts = []
+        for posting in transaction.postings:
+            if posting.meta is None or UNCONFIRMED_ACCOUNT_KEY not in posting.meta:
+                continue
+            group_number = existing_groups.setdefault(posting.account,
+                                                        num_groups)
+            predicted_account_names.append(posting.account)
+            if group_number == num_groups:
+                num_groups += 1
+            group_numbers.append(group_number)
+        return predicted_account_names, group_numbers
+
     def _get_unknown_account_predictions(self,
                                          transaction: Transaction) -> List[str]:
-        group_prediction_inputs = self._feature_extractor.extract_unknown_account_group_features(
-            transaction)
-        group_predictions = [
-            self.predict_account(prediction_input)
-            for prediction_input in group_prediction_inputs
-        ]
-        group_numbers = training.get_unknown_account_group_numbers(transaction)
-        return [
-            group_predictions[group_number] for group_number in group_numbers
-        ]
+        if self._has_unconfirmed_account(transaction): 
+            # if any of the postings have an unconfirmed account, then prediction was handled by smart_importer
+            predicted_account_names, _ = _group_predicted_accounts_by_name(transaction)
+            transaction.meta[UNCONFIRMED_ACCOUNT_KEY] = True
+            return predicted_account_names
+        else:
+            group_prediction_inputs = self._feature_extractor.extract_unknown_account_group_features(
+                transaction)
+            group_predictions = [
+                self.predict_account(prediction_input)
+                for prediction_input in group_prediction_inputs
+            ]
+            group_numbers = training.get_unknown_account_group_numbers(transaction)
+            return [
+                group_predictions[group_number] for group_number in group_numbers
+            ]
 
     def _make_candidate_with_substitutions(self,
                                            transaction: Transaction,
@@ -754,8 +791,12 @@ def _make_candidate_with_substitutions(self,
             unique_id: account
             for unique_id, account in zip(unique_ids, new_accounts)
         }
-        group_numbers = training.get_unknown_account_group_numbers(transaction)
-        unknown_names = training.get_unknown_account_names(transaction)
+        if transaction.meta is not None and UNCONFIRMED_ACCOUNT_KEY in transaction.meta:
+            _, group_numbers = self._group_predicted_accounts_by_name(transaction)
+            unknown_names = self._get_unknown_account_names(transaction)
+        else:
+            group_numbers = training.get_unknown_account_group_numbers(transaction)
+            unknown_names = training.get_unknown_account_names(transaction)
         substitutions = [
             AccountSubstitution(
                 unique_name=unique_id,