The aim of this comment is to discuss two issues:

1. how Claro  was analysed, once it was considered to be an ellipsis;
2. what is the form / function of paciência  in the sentence.
 

1. The above sentence presents one ellipsis and possibly a second one (cf. 2 a)).

The first ellipsis to take into account is present in:

Claro que preferia correr em condições idênticas....

The analyis given by the parser was the following:

Claro    [claro] ADV@ADVL
que    [que] KS @SUB @#FS-A<
preferia (...)
correr    (...)
 

Another possibility is to reconstruct a possible ellipsis that is present here. And in this case, Claro would be, instead of an @ADVL, the subject complement (@SC) of the elliptic verb ser :

(É) claro que preferia correr.....

And consequently, the clause initiated by the subordinate conjunction would be the subject (@SUBJ):

Claro    [claro] ADJ M S @SC
que    [que] KS @SUB @#FS-<SUBJ
preferia (...)
correr    (...)

The question here is to determine the morphological class of the word. As claro may be an adverb or an adjective, then the syntactic analysis may depend on that, either considering an ellipsis and therefore choosing to make it visible (attributing the syntactic function of subject complement) or an ADVL (~ claramente)
 

Sentence C170-12 is a very good example of the choices taken when dealing with ellipses.
 
 

2. Let's concentrate on the period: não sendo possível, paciência.

Three possibilities were taken into account:

a) paciência as an averbal clause

This possibility would mean that there is an ellipsis present in the period:

...não sendo possível, (há que haver) paciência

                 or

...não sendo possível, (temos de ter) paciência

This case would fit perfectly in the whole sentence itself, since we would then have  a compound unit with two clauses (the first conjoint being a finite clause, and the second conjoint an averbal clause). Syntactically, and bearing in mind the reconstruction of the ellipsis, paciência would be the direct object of the elliptic verb: ter/ haver.
However, the ellipsis is not immediately reconstructed by the reader, in part because there is not a unique possibility (as it can be seen above) and also because there are no co-textual elements, previously in the sentence, that may help that reconstruction.
 

Furthermore, in similar cases like azar, the reconstruction of the hypothetic ellipsis would be different:
 

Ex: Não conseguimos financiamento, azar .( ~ foi azar)
                                                                               ( ~ tivemos azar)
                                                                               ( ~ por azar)
 

Let's see some examples taken from the CETEMPúblico corpus:
 

Ext 5806 (opi, 98b): Na Alemanha, onde a OCDE estima para este ano um crescimento de 2,7 por cento, o desemprego praticamente não diminui, nem se prevê que nos tempos mais próximos -- azar de Kohl...  (source: CETEMPúblico)
 

Ext 18293 (soc, 95b): Por isso, é natural, nas letras rap, ouvirem-se histórias como a daquele negro que se «fez à life» e -- azar -- foi «de cana» .  (source: CETEMPúblico)
 

Ext 268141 (soc, 95a): Trata-se de um empresário cinquentão que , azar, teve de andar uns tempos com o carro da filha . (source: CETEMPúblico)

Ext 1076000 (des, 96a): «Se não quiserem ver-me jogar, azar. (source: CETEMPúblico)
 

b) Noun

Because of the lack of regularity seen while trying to reconstruct the ellipsis, and in this case, having already the syntactic function (conjoint), the option taken was to consider paciência a noun.
However, this case might not work when the sentence is not a compound unit. Let's take the last example in a):

Ext 1076000 (des, 96a): «Se não quiserem ver-me jogar, azar. (source: CETEMPúblico)

Here we have an adverbial clause (conditional clause), therefore we would expect to have a main clause. Instead, and not considering azar as part of an elliptic clause, we have a noun and determining its syntactic function is a problem.

c) Interjection

And finally, in terms of form, paciência or similar words (azar, for instance) could be considered as being interjections. However the same problem as to determining the syntactic function would remain the same, if the sentence is not a compound unit.
However, many interjections belong to other morphological classes and are used in an "interjectional way" (Bick, 2000).
This is also the case of paciência, a noun possibly used in this context in an "interjectional way". The same happening to azar. So, the original word class should be mantained.